Kantian Humility. Our Ignorance of Things in Themselves 9780198236535, 0198236530

Table of contents :
Contents......Page 10
Note on Sources and Abbreviations......Page 14
Introduction......Page 16
1. Introduction......Page 22
2. Allison's Deflationary Proposal......Page 23
3. Reasons for Suspicion......Page 25
4. A Metaphysical Proposal, and an Acid Test Passed......Page 27
1. Introduction......Page 30
2. Bennett on Two Distinctions and their Conflation......Page 39
3. A Case For, and Against, the 'Bare Substratum'......Page 43
4. Three Theses: Some Further Texts......Page 48
5. The Distinction......Page 49
6. Humility......Page 56
7. Receptivity......Page 58
2. The Pure Concept of Substance vs. the Schematized......Page 63
3. The Concept of Phenomenal Substance in General......Page 68
4. Matter as a Merely Comparative Subject......Page 71
5. Note on the Inference to Substance......Page 79
6. Concluding Remarks and New Business......Page 81
1. Introduction......Page 83
2. Kant's Version of Leibniz......Page 85
3. A Distinction between Phenomena and Things in Themselves......Page 89
4. A Reduction of Phenomena to Things in Themselves......Page 93
5. Knowledge, via Phenomena, of Things in Themselves......Page 104
6. Kant and Leibniz on Relations......Page 108
1. An Early Distinction between Phenomena and Things in Themselves......Page 112
2. The Principle of Succession, and Receptivity......Page 119
3. The Principle of Coexistence, and Irreducibility......Page 122
4. Analysis of the Argument for Irreducibility: Preliminaries......Page 124
5. Irreducibility Argument I......Page 127
6. Irreducibility Argument II......Page 130
7. Concluding Remarks......Page 136
1. Assembly......Page 139
2. An Imaginative Exercise......Page 140
3. Kant, Leibniz, and a Mirror Broken......Page 142
4. Later Signs of the Irreducibility Argument......Page 146
1. A Phenomenalist Reading of the Comparison......Page 155
2. Problems, and a Contradiction......Page 157
3. A Different Lockean Distinction......Page 162
4. A Contradiction Dissolved......Page 170
5. A Closer Look......Page 173
1. Introduction......Page 177
2. Bennett's Instructive Mistake......Page 178
3. Spatial Features and Space-Filling Features......Page 180
4. A Caveat about Space......Page 181
5. Kant's 'Primary' Qualities: Geometrical and Dynamical......Page 183
6. Solidity vs. Impenetrability, and a Problem for a Contemporary Orthodoxy......Page 187
7. Science: Primary vs. Tertiary Qualities......Page 192
8. Objectivity: Primary vs. Tertiary Qualities......Page 197
1. Experience and the Unobservable......Page 201
2. The Kant–Eberhard Controversy: Unobservable vs. Supersensible......Page 206
3. Observability and Community......Page 210
4. Monadology, Well vs. Badly Understood......Page 212
5. Final Comments......Page 218
1. Summing Up......Page 220
2. Idealism: First Impressions......Page 221
3. Idealism: Things in Themselves......Page 222
4. Idealism: Space......Page 225
Bibliography......Page 234
D......Page 242
H......Page 243
M......Page 244
R......Page 245
S......Page 246
W......Page 247

Citation preview

KANTIAN HUMILITY

This page intentionally left blank

Kantian Humility Our Ignorance of Things in Themselves

RAE LANGTON

CLARENDON PRESS • OXFORD

Tfe boofc has been printed digitally and produced in a standard speaftaticm In wder to ensure its continuing availability

OXFORD UWrVJB&SITY PRESS

Great Clarendon Street, Oxford OX2 6DP Oxford University Press is a department of the University of Oxford. It furthers the University's objective of excellence in research,, scholarship, and education by publishing worldwide in Oxford New York Auckland Cape Town Dar es Salaam Hong Kong Karachi Kuala Lumpur Madrid Melbourne Mexico City Nairobi New Delhi Shanghai Taipei Toronto With offices in Argentina Austria Brazil Chile Czech Republic France Greece Guatemala Hungary Italy Japan South Korea Poland Portugal

Singapore Switzerland Thailand Turkey Ukraine Vietnam

Oxford is a registered trade mark of Oxford University Press in the UK and in certain other countries Published in the United States by Oxford University Press Inc., New York © Rae Langton 1998 The moral rights of the author have been, asserted Database right Oxford University Press (maker) Reprinted 2007 All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, without the prior permission in writing of Oxford University Press, or as expressly permitted by law, or under terms agreed with the appropriate reprographics rights organization. Enquiries concerning reproduction outside the scope of the above should be sent to the Rights Department, Oxford University Press, at the address above You must not circulate this book in any other binding or cover And you must impose this same condition on any acquirer ISBN 978-0-19-823653-5

Wir sehen dm Innere der Dinge gar nickt em,

We have no insight whatsoever into the intrinsic nature of things. (A277/B333)

This page intentionally left blank

ACKNOWLEDGEMENTS Thanks are clue to many friends, teachers, and colleagues for patient and constructive comments on earlier thoughts and drafts, without which this book would be much the poorer, The remaining errors are, needless to say, all my own work. The earliest inklings of this story about Kant came to me in Sydney in 1985, encouraged by Keith Campbell (who first inspired my interest in Kant) and David Stove (who warned me that Kant was a Black Hole—encouragement enough for the countersuggestible). Later on, at Princeton and thereafter, I was lucky to have had the advice of Margaret Wilson, who supervised the dissertation which was ancestor to this book. Her patience and thoroughness saved me from many errors and overglib generalizations. I was fortunate to receive comments from others, including Mark Johnston, Eckart Forster, Michael Friedman, and Beatrice Longuenesse. Conversations with Alison Lay wine were also very helpful. Since my return to Australia in 1990, the Philosophy Department at Monash University, and the Philosophy Program at the Research School of Social Sciences, Australian National University, have each provided me with the great luxury of teaching-free time, a semester's leave at Monash in 1994, and a Fellowship at the Research School which began in 1997,1 have benefited enormously from interaction with my colleagues at both institutions, especially John Bigelow, Jamie Dreier, Frank Jackson, and Michael Smith. Without Lloyd Humberstone and David Lewis I would have been muddled—or more muddled—about the metaphysics of intrinsic properties, Simon Blackburn has been particularly helpful on the subject of powers and unknowable somethings. I have enjoyed conversations with Michael Avers about Kantian epistemic humility and its Lockean analogues. I am grateful for the detailed and constructive advice I received from the readers of Oxford University Press, Robert Stern and Peter Strawson. The book has been much improved by the wise editorial hand of Angela Blackburn. Richard Holton has helped philosophically and practically in ways that I cannot even begin to count. Eleanor Holton, whose impending arrival spurred this book to completion, has helped in a very practical way. This book is dedicated to the memory of my father, David Langton, who once said I should write a book; though I somehow doubt he had this one in mind.

This page intentionally left blank

CONTENTS Note on Sou-reef and Abbreviations

xiii

Introduction

i

I. AN OLD PROBLEM

7

1. 2. 3. 4.

Introduction Allison's Deflationary Proposal Reasons for Suspicion A Metaphysical Proposal, and an Acid lest Passed

7 8 10 12

2. THREE KANTIAN THESES

15

1. 2. 3. 4. 5. 6. 7.

15 24 28 33 34 41: 43

Introduction Bennett on Two Distinctions and their Conflation A Case For,, and Against, the 'Bare Substratum' Three Theses: Some Further Texts The Distinction Humility Receptivity

3. SUBSTANCE AND PHENOMENAL SUBSTANCE

48

1. 2. 3, 4, 5, 6.

48 53 56 64 66

Introduction The Pure Concept of Substance vs. the Schematized The Concept of Phenomenal Substance in General Matter as a Merely Comparative Subject Note on the Inference to Substance Concluding Remarks and New Business

4, LEIBNIZ AND KANT

68

1. 2. 3. 4. 5. 6.

68 70 74 78 89 93

Introduction Kant's Version of Leibniz A Distinction between Phenomena and Things in Themselves A Reduction of Phenomena to Things in Themselves Knowledge, via Phenomena, of Things in Themselves Kant and Leibniz on Relations

x Contents 5. KANT'S REJECTION OF REDUCTIBILITY

97

1. An Early Distinction between Phenomena and Things in Themselves 2. The Principle of Succession, and Receptivity 3. The Principle of Coexistence, and Irredueibility 4. Analysis of the Argument for Irredueibility: Preliminaries 5. Irredueibility Argument I 6. Irredueibility Argument II 7. Concluding Remarks

97 104 107 109 112 115 121

6. FITTING THE PIECES TOGETHER

124

1. 2. 3. 4.

124 125 127 131

Assembly An Imaginative Exercise Kant, Leibniz, and a Mirror Broken Later Signs of the Irredueibility Argument

7. A COMPARISON WITH LOCKE

140

1. 2. 3. 4. 5.

140 142 147 155 158

A Phenomenalist Reading of the Comparison Problems, and a Contradiction A Different Lockean Distinction A Contradiction Dissolved A Closer Look

8. KANT'S 'PRIMARY' QUALITIES

162

1. 2. 3. 4. 5. 6.

162 163 165 166 168

Introduction Bennett's Instructive Mistake Spatial Features and Space-Filling Features A Caveat about Space Kant's 'Primary' Qualities: Geometrical and Dynamical Solidity vs. Impenetrability, and a Problem for a Contemporary Orthodoxy 7. Science: Primary vs. Tertiary Qualities 8. Objectivity: Primary vs. Tertiary Qualities

172 177 182

9. THE UNOBSERVABLE AND THE SUPERSENSIBLE

186

1. 2. 3. 4. 5.

186 191 195 197 203

Experience and the Unobservable The Kant-Eberhard Controversy: Unobservable vs. Supersensible Observability and Community Monadology, Well vs. Badly Understood Final Comments

Contents

xi

10. REALISM OR IDEALISM?

205

1. 2. 3. 4.

205 206 207 210

Summing Up Idealism: First Impressions Idealism: Things in Themselves Idealism: Space

Bibliography

219

Index

227

This page intentionally left blank

NOTE ON SOURCES AND ABBREVIATIONS References to Kant's Critique of Pure Reason follow the customary practice of citing the pagination of the 1781 (A) edition and the 1787 (B) edition. Translations of the Critique are chiefly my own, following the German edition by Raymund Schmidt, though influenced by Norman Kemp Smith's translation. Citations of other writings by Kant give their location in the Academy edition of Kant's works, and (usually) an English translation. Unless otherwise indicated, quotations from the Prolegomena, the Melaphysical Foundations of Natural Science, the Letters, and On a Discovery follow the translations by Peter Lucas, J. Ellington, Arnulf Zvveig, and Henry Allison, respectively. Translations from all other works by Kant are my own, though influenced by the English translations given in the Bibliography. Kant's Reflexionen zur Metaphysik are cited by the numbers assigned to them in the Academy edition; dates given are those suggested by Erich Adickes (in the Preface to volume xvii of the Academy edition), or by Arthur Melnick (in Space, Time and- Thought in Kant, 547—9), signified by the letters 'A' and 'M' respectively, In addition the following abbreviations are used for works by Kant (full publication details can be found in the Bibliography): Ak, Allison Beck Ellington Inaugural Dissertation Kerferd and Walford Living Forces Lucas Metaphysical Foundations New Exposition

Gesammelte Schriften, Academy edition The Kant-Eberhard Controversy, trans. Henry Allison Kant's Latin Writings, trans. L. W. Beck et al. Metaphysical Foundations of Natural Science, trans. J. Ellington De mundi sensibilis ataue intelligibilis forma et principiis (1770) Selected Pre-Critical Writings, trans. G. B. Kerferd and D. E. Walford Gedanken von der wahren Schatzung der lebendige Krafte (1747) Prolegomena, trans. Peter G. Lucas Metaphysische Anfangsgrunde der Naturrpissenschaft (1786) Principiorum primorum cognitionis tnetaphysicae nova dilucidatio (1755)

xiv

Note on Sources, and Abbreviations

On a Discovery Physical Monadology R Regions in Space What Real Progress?

Zweig

Uber eine Entdeckung nach der alle neue Kritik der reinen Vernunft (lurch erne iiltere entbehrlich gernacht werden soil (1790) Metaphysicae cum geometria iumtae mm in philosophia naturali, cams specimen I. conlinet mona-dologican physicam (1756) Reftexionen zur Metaphysik Von dent ersten Grunde des Unterschiedes der Gegenden in Raume (1768) Welches sind die ivirklichen Fortschritte, die die Metaphysik sett Leibnizem und Wolf's Zeiten in Deutschland gemacht hat? (1791/1804) Kant's Philosophical Correspondence 1759—99, trans. Arnulf Zweig.

Works by Leibniz are cited by original language source and English translation. Unless otherwise indicated, quotations of Leibniz follow the transi abbreviations are used for works bv Leibniz: lations cited. The following &' Anew and Garber Bodemann Couturat Gerhardt Gerhardt M Grua Locmker Remnant and Bennett

Philosophical Essays, trans. R. Ariew and D. Garber Die Leibniz-Handschriften der koniglichen ojfentlichen Bibliothek zu Hannover, ed. E. Bodemann Opuscules et fragments inedits de Leibniz, ed. L. Couturat Die phihsophischen Schriften, ed. C. I. Gerhardt Mathematische Schriften, ed. C. I. Gerhardt Textes inedits, d'afres les manuscnts de la biblioiheque provinciate de Hanovre, ed. G. Grua Philosophical Papers and Letters, trans. L. Loemker New Essays on Human Understanding (1765), trans. P. Remnant and J. Bennett.

INTRODUCTION Many historians of philosophy, with all their intended praise, let the philosophers speak mere nonsense. They do not guess the purpose of the philosophers . . . They cannot see beyond what the philosophers actually said, to what they really meant to say. On a Discovery (1790)

Kant is rather scathing of a certain kind of history of philosophy. Alert to the possibility that interpretive charity and interpretive fidelity may sometimes pull In different directions, he favourably contrasts the stance of intelligent and charitable engagement with the stance of an uncritical devotee. If Kant is right to value the former, then the history of philosophy since Ms time has been fortunate, since there has been no lack of intelligent and charitable engagement. This is perhaps especially true with respect to interpretations of Kant himself. What did Kant himself actually say; or mean to say? He said that there is a distinction between appearances and things in themselves, that things in themselves exist, and that we have no knowledge of things in themselves. This has seemed to come dangerously close to nonsense, and there has accordingly been no lack of charitable interpreters only too willing to suppose that he did not mean quite what he said, and that a different purpose must be guessed. Kant's purpose must be to say something else, something mild, and sane, and wise, Kant gives some licence, as we can see, for this kind of rational reconstruction: he gives licence for seeing beyond what the philosopher actually said, to what he really meant to say, It seems unlikely, though, that a philosopher did not mean to say what he said, when he said it over and over again. And it is worth bearing in mind that Kant himself did not appear to think his philosophy was mild, and sane, and wise. Startling, yes; frustrating, yes; wise, perhaps; mild, no.1 Faced with this stubborn fact, there is a choice. One can let the philosopher speak nonsense. One can turn an indulgent blind eye to the metaphysical lapses. Or, best of all, one can aim for an alternative which achieves both fidelity and charity, which accepts that this is what the philosopher actually said, and that it is what the philosopher meant to say, and that it is not nonsense after all. 1

Kant says it is 'startling' ^285/6341), and admits that it leaves unsatisfied an 'inextinguishable yearning' (A796/B824). An example of a sane and mild interpretation will be discussed in Chapter i,

2

Introduction

Kant says that there Is a distinction between things in themselves and phenomena. What he says is admittedly metaphysics, but none the worse, I think, for that. He says that things as we know them consist 'wholly of relations' (A285/B34i). He says that we have no insight into 'the inner' of things (A277/11233), He has a distinction between substances, bearers of intrinsic properties, on the one hand; and relational properties of substances on the other. He says that we have no knowledge of the intrinsic properties of substances. This, as it stands, is not idealism, but a kind of epistemic humility. There are inevitable constraints on what we can know, inevitable limits on what we can become acquainted with. And while those limits could be correctly described, in Strawson's phrase, as 'the bounds of sense',, such a description fails to capture Kant's thought, that there is a particular sort of thing that is beyond the bounds of sense, something abstractly charaeterizable in metaphysical rather than episternological terms: not simply as 4 that which is beyond the bounds of sense', but as 'that which has an intrinsic nature', I introduce this way of understanding Kant by suggesting that it can help to dissolve a very old contradiction: things in themselves exist, and are the causes of phenomena, and we have no knowledge of things as they are in themselves. This problem has been described as presenting the 'acid test* for any interpretation of things in themselves,2 The problem and my suggested solution is the topic of Chapter i. In Chapter 21 state and explain in more detail three of the fundamental Kantian theses which will be the focus of my attention. The first is the distinction between phenomena and things in themselves. The second is Kant's denial of knowledge of things as they are in themselves, his thesis of epistemic humility. And the third is the empiricist strand in Kant's philosophy, his belief that we are receptive creatures, who must be affected by the things of which we come to have knowledge, Kant believes that humility follows from this fact of receptivity: he believes, as Strawson has remarked,3 that our ignorance of things as they are in themselves follows from the fact that we must be affected by things if we are to achieve knowledge of them. If this is correct, then our ignorance of things as they are in themselves is not supposed to be a special consequence of the arguments about space, or time, or the categories: it is supposed to be a general consequence of the fact that human knowledge is receptive. This has two implications. The first is dialectical. Many philosophers would agree with Kant about the receptivity of human knowledge; many - Henry Allison, Kant's Transcendental Idealism (New Haven, Conn.: Yale University Press, 1.983), 247. • F, K Strawson, 1'he Htmnds of Seme (\Amdvn: Methuen, 1966), 250.

Introduction 3 who would raise an eyebrow at other doctrines of the Critique will find common ground here. But if there is anything to Kant's belief that humility follows from receptivity, then this poses a rather immediate threat to ordinary epistemological ambition. The premise which Is supposed to lead us to humility is true, or widely accepted to be true, If Kant is right, then many philosophers are closer than they think to the Kantian conclusion that we have no knowledge of things as they are in themselves. The second Implication is exegetical. Suppose we are interested in the question of why Kant believes we have no knowledge of things as they are in themselves. If humility is supposed to follow from receptivity, then it should be possible to explore this question without exploring In detail the arguments about space, time, and the categories, for which Kant is (perhaps justly) most famous. The arguments from the Aesthetic and Analytic are accordingly given little detailed attention in the following discussion, since, notwithstanding their importance, they are separable from the conclusion about our ignorance, However, there is an obvious problem, and it is raised at the end of Chapter 2. Although Kant believes that humility follows from receptivity, it does not appear to do so. Strawson calls this a fundamental unargued premise of the Critique. From the fact that we must be affected if we are to have knowledge it does not directly follow that we must be ignorant of things In themselves—or not at any rate without some further premise. The search for that missing premise is the task of Chapters 4 to 6. Before it can be pursued, there is a potential stumbling block to be removed. According to Kant's distinction, as I interpret him, things in themselves are substances. What then of Kant's claim in the Critique that substance, phenomenal substance, is in the world that we experience? In Chapter 3 T explain how this commitment to phenomenal substance should be understood, in light of Kant's basic distinction. The problem raised at the end of Chapter 2 Is taken up in Chapter 4: the problem of how epistemic humility is supposed to be a consequence of receptivity. Some of the clearest expressions of Kantian humility are to be found in the context of Kant's critique of Leibnizian philosophy. This suggests a clue—a possible path to a solution, a possible link between Kant's attitude to Leibnk and his belief in our ignorance of things as they are in themselves. So in this fourth chapter I offer an interpretation of Leibniz that attempts to show what, in Kant's opinion, Leibniz was doing-—and what he was doing wrong, There are four notable features to Leibniz's philosophy, or at least to Kant's version of it. There is a metaphysical distinction between things in themselves and phenomena, which coincides with a distinction between substances and their relational properties. There is a commitment to the

4 Introduction reducibility of relations. There Is a denial of causation, and hence a denial of sensory receptivity. And finally, there is an extraordinary ambition that is quite the antithesis of epistemic humility. Leibniz believes that there are no bounds of sense, and that we are each acquainted, through the senses, with everything there is, although our sensory acquaintance is confused. To what extent do Kant and Leibniz share common ground? They agree about certain metaphysical theses: in particular about the distinction between phenomena and things in themselves. They disagree about epistemology: about receptivity, and humility, Kant says, contrary to Leibniz, that human knowledge is receptive, and that we have no knowledge of things as they are in themselves. A question mark stands, though, over the metaphysics of the reducibility of relations. Is this a point of similarity between the two philosophers, or a point of difference? Some critics affirm the former, Some even suggest that a Leibnizian view about relations led Kant to his mature philosophy. If they were right, then Kant's attitude to Leibniz would indeed provide a clue as to why he believes we have no knowledge of things as they are in themselves. To discover whether they are right, and what Kant's attitude to Leibniz has to do with the thesis of humility, I turn in Chapter 5 to the New Exposition (1755). Two of the above Leibnizian themes are addressed explicitly in this early work. There is an explicit distinction between phenomena and things in themselves which has much in common with that of Leibniz. Kant's arguments hinge on a contrast between how a substance is when it is by itself, and how it is when it is in a relation to other things: a contrast between the intrinsic and relational properties of a substance. Kant's matter theory—his theory of force—is intricately connected with this basic distinction between phenomena and things in themselves, and this early work helps to show how. Moreover, there is an explicit focus here on the issue of the reducibility of relations. Kant's argument is of considerable interest and sophistication, and deserves close attention. However, contrary to the suggestions of the aforementioned critics, it does not endorse a Leibnizian view about relations. Its conclusion is that relations, or relational properties, are irreducible. Kant believes that relational properties, causal powers in particular, are not reducible to intrinsic properties-—and he seems to believe that intrinsic properties are therefore causally inert, lie says that it is never through its own intrinsic properties that a substance has the capacity to affect other things, On the two metaphysical themes there is thus partial agreement and partial disagreement with Leibniz: agreement about the basic distinction between phenomena and things in themselves, disagreement about the reducibility of relations. The arguments of this work also have implications

Introduction 5 for the two epistemological themes, one drawn by Kant, the other not—or at least, not yet. There is, already, the theme of receptivity. The New Exposition, notwithstanding its status as an essay in dogmatic metaphysics,, contains a seed of empiricism: it offers an anti-Leibnizian argument which implies that human knowledge is essentially receptive, and Kant explicitly draws this conclusion. Most important of all, there is an apparent implication—which Kant does not yet draw—for episternic humility. The conclusion about relations, together with the conclusion about receptivity, has the potential to yield the conclusion about humility. If this is so, then Kant's attitude to Leibniz does indeed provide a clue as to why he believes we have no know-ledge of things as they are in themselves: but, contrary to the critics, the clue is to be found not in his agreement with. Leibniz about the metaphysics of relations, but rather in his disagreement. How these pieces fit together is the topic of Chapter 6. Receptivity implies that we can have knowledge only of what can affect us. Irreducibility implies, in Kant's view, that intrinsic properties cannot affect us. If Kant maintains this view throughout his philosophical career—as I shall argue he does—then one can trace a path to humility. If substances affect us, but it is not through their intrinsic properties that we are affected, then their intrinsic properties remain unknown. One might put the argument in terms of a familiar Leibnizian metaphor. If perception were to mirror bodies, and bodies were to mirror the intrinsic properties of monadic substances, then perception would mirror the intrinsic properties of substances. Our minds would, through perception, mirror the monadic realm of things in themselves. The intrinsic properties of substances would be known through perception (albeit confusedly). The epistemological ambition of Leibniz's philosophy—his denial of humility—arguably depends on this commitment to the reductibility of relations. But if, as Kant says, the relations that constitute bodies are not reducible, then bodies do not after all mirror the intrinsic properties of substances, The mirror is broken. Perception cannot mirror the intrinsic properties of substances. Their intrinsic properties must remain unknown. Chapters 7 to 9 explore the implications of this understanding of Kant for an otherwise surprising feature of his philosophy, namely his scientific realism. One aspect of this realism is an apparent commitment to a distinction between primary and secondary qualities that applies within the phenomenal world. This is surprising in a supposed idealist, and appears to conflict with Kant's apparently Berkeleyan claim in the Prolegomena, that he makes all the qualities secondary. In Chapter 7 I resolve the conflict: Kant does not mean that he makes the qualities ideas, but that he makes them powers. The result is that Kant has a version of the primary/secondary

6 Introduction quality distinction that is novel, interesting, and has considerable advantages over many of its competitors, whether of his own time or of ours. Kant's version of this distinction, and its merits, is the topic of Chapter 8. Kant's argument in the Critical period against a more traditional view of primary qualities draws on. the early irreducibility argument of the New Exposition. It has the potential to challenge some of our own contemporary philosophical orthodoxies, or so it seems to me. Chapter 9 turns to a further aspect of Kant's scientific realism, namely his commitment to the unobservables of science. This commitment is permitted by his understanding of receptivity: we can have knowledge of anything that can affect us, Kant's scientific realism, and his humility, thus have a common origin. Does the interpretation. I offer in these pages amount to a rational reconstruction of Kant-—is it an attempt to see beyond what the philosopher actually said, to what he really meant to say? It is my sincere hope and belief that the answer to this question is no. The interpretation aims to be firmly grounded in what the philosopher actually said. However., it does ascribe to a notorious idealist a position that is not idealism, not anti-realism of any kind, but rather epistemic humility. It does make a metaphysician of a philosopher who is supposed to have abandoned metaphysics. And it does ascribe to a philosopher of long ago a view that has contemporary resonance: many philosophers today are indeed closer than they think to the Kantian conclusion that we have no knowledge of things as they are in themselves. There is a gulf between Kant the notorious idealist, and the realist Kant who emerges in the pages that follow, but that gulf can sometimes be explained. Many apparent expressions of idealism turn out to be expressions of epistemic humility—many, but admittedly not all. Many apparent expressions of idealism rest on dubious interpretive assumptions about what phenomena must be—many, but admittedly not all. Some qualifications and caveats are called for, and these are raised in the final chapter, They complicate the story, but they do not, I believe, undermine it. It remains as the story of what I find to be the central stream of Kant's thinking—or of what I hope others may find to be, at the very least, one driving current within it.

I

An Old Problem I. Introduction Kant affirms the existence of things in themselves and speaks of them affecting our minds, and being the cause of appearances. Since the earliest days, this has been seen as the fundamental sticking point in Kant's philosophy.1 If Kant's philosophy is right, then we have no knowledge of things in themselves: we cannot know that they exist, nor can we know that they are causes of appearances, and affect us.2 Kant's philosophy has been thought to imply two relations of affection, empirical and transcendental. We are affected by bodies; and we are affected by things in themselves. Both relations of affection seem illegitimate on the assumption of idealism, since the first requires causal agency of a mere representation, and the second is supposed to be unknown. Our concern here will be chiefly with the problem raised by things in. themselves. The problem, thus described, attributes to Kant two metaphysical theses. Ki Things in themselves exist, K.2 Things in themselves are the causes of phenomenal appearances, And it attributes to Kant an episteoiologieal thesis. K.3 We can have no knowledge of things in themselves. Trouble comes with the conjunction of the three. For the epistemological thesis appears to imply these corollaries: 1 Jacob! being one of the first to draw attention to it; F, H. Jacobi, Werke (Leipzig: Gerhard Fleischer, 1.8:15), "• 3°4- See Allison's discussion of this problem, Kant's Transcendental Idealism, 247-54, Paul. Guyer shares my suspicion of what, he calk Allison's 'anodyne conceptual analysis', though he does not give the same reasons for suspicion. He too thinks that Allison's attempted, solution to the problem fails, See Guyer, Kant and the Claims of Knowledge (Cambridge: Cambridge University Press, 1987), ch. 15, especially p, 338, My explanation of Allison's proposal draws partly on Guyer's helpful discussion, though he would disagree with my proposed solution. 2 The 'double affection' interpretation of Kant is developed by Erich Adickes, who (oddly) sees double affection not as a problem but as a solution. See Kanls Lehre von der doppclien Affektion unseres I eh ah Scftlussel zu seiner Erkenntmsthcmie (Tubingen: J, (1 Molir, 1:929).

8

An Old Problem Ci We cannot know that things in themselves exist. Ca We cannot know that things in themselves are the causes of phenomenal appearances.

We cannot know Ki and K2. Kant's storv makes itself imtellible. 2, Allison's Deflationary Proposal In Chapter n of Kant's Transcendental Idealism, Henry Allison confronts these problems. His response to the alleged problem of empirical causation is swift. It is a mistake, he says, to suppose that Kant cannot legitimately speak of affection by empirical objects; 'Kant not only can but does speak about the mind as affected by empirical objects.*3 He gives exa.m.ples. But Allison cannot argue by equivocation—even if he not only can but does. Allison's 'can' is equivocatory. That Kant does something implies that he can do something in one sense, but not in the other. As Kant reminds us, should reminders be needed, doing and legitimately doing are not the same.4 There is more,, though, to Allison's defence. He says, and I think rightly, that a problem of empirical affection would arise only for a Berkeley an Kant, for whom appearances are just ideas, and he says that the Berkeleyan Kant is a fiction. And he saves his most elegant solution for the most fundamental problem, the one that concerns us here: namely, the existence of, and affection by, unknowable things in themselves. The mistake, says Allison, is to construe Kant's basic theses as metaphysical theses. Construed metaphysically, the contrast between phenomenal appearance and thing in itself lands us with 'two distinct entities' or 'two kinds of entity'.5 That is where the trouble starts, and that is where we can stop it in its tracks. Kant is not interested in making existence claims, says Allison. He is interested in philosophical methodology. It is not that there exist two kinds of thing, phenomena and things in themselves. Rather, there are two ways of considering things, We can consider things, at the empirical level, in relation to our sensibility. And we can consider things, at the transcendental level, in abstraction from that relation. When doing science we sometimes consider a thing in abstraction from certain properties it has,

3

Allison, Kant '$ Transcendental Idealism, 249, emphasis added. My objection is not to the overall strategy of his defence of empirical affection, but to this particular punning argument. 4 Kant distinguishes the question of right (quidjuris) from the question of fact (quidfacti) at the beginning of the Transcendental Deduction (A84/Bn6). 5 .Allison., Kant's Transcendental Idealism, 340, 248,

An Old Problem 9 such as weight: but this does not show that there are weightless things. When evaluating an applicant for a job, we sometimes consider the applicant in abstraction from certain properties she has, such as height; but this does not show that there are people who have no height. When doing philosophy, we sometimes consider things in abstraction, from their relation to our sensibility, in abstraction from their spatial, temporal, categorial properties; but this does not show that there are non-spatial, atemporal, non-causal things. Kant's distinction between phenomena and things in themselves is a distinction between two ways we have of considering things,, and from that distinction no metaphysical theses follow. In the absence of metaphysical theses, no further problems arise. In place of the metaphysical Ki and K.2, we have the following anodyne theses; A i We can consider things 'in themselves', i.e. in abstraction from the conditions of our sensibility. A.2 Things considered in abstraction from the conditions of our sensibility can be considered only as something that affects the mind. Allison's view makes it analytic that things in themselves are not describable spatioternporally. Statements about things in themselves are, by definition, statements that abstract from any talk of space, time, and the categories. This kind of abstraction is just what constitutes the transcendental level of considering things. Since knowledge arises only with the concrete application of the forms and categories, Kant's thesis K3 about our ignorance of things in themselves becomes nothing *& more than this: A3 Things considered in abstraction from their relation to our sensibility are things considered in abstraction from their relation to our sensibility. On Allison's view, there are no unknowable entities in the picture. If we are asked to say more about the 'something' in A.2 that is supposed to affect the mind, we must refuse. To say anything more at the transcendental level would be to make oneself guilty of a kind of methodological impurity. We can say nothing more until we stop abstracting, until we descend to the empirical level, Then, of course, our answer to the question 'What affects the mind?' must be an empirical answer. Light, air, elements, and all the familiar denizens of the globe are the things that affect our minds. There is no problem of affection. The 'something' in A2 is not something over and above the familiar phenomenal objects: it is identical with the class of phenomenal objects 'referred to collectively'. It is the class of those objects, considered in an abstract way.

io

An Old Problem 3. Reasons for Suspicion

Allison's is an ingenious and attractive solution to an old and ugly problem. But I would like to suggest that there are reasons for suspicion. I have an ulterior motive. I have up my sleeve a solution, that is> though less ingenious, more attractive. First, there is a problem about analyticity, Allison's approach makes it analytic that we have no knowledge of things in themselves. To consider things in themselves is simply to consider things in abstraction from the conditions of our knowledge: K.3 has become the tautological Aj. From one point of view this is an advantage, but from another it is a grave defect, for it foils to do justice to an aspect of Kant that ought not to be ignored. What 1 have in mind is not exactly a Kantian, philosophical thesis,, in the sense that Ki—K.3 are philosophical theses. Rather, it is a Kantian attitude to these philosophical theses, and in particular to the third. When Kant tells us that we have no knowledge of things in themselves, he thinks he is telling us something new and important. The truth of K3 is a major philosophical discovery. Moreover, it is not just a discovery with a definite, non-trivial content. It is a depressing discovery. Kant thinks we are missing out on something in not knowing things as they are in themselves. Kant speaks of our yearning for something more, he speaks of doomed aspirations, he speaks of 'our inextinguishable desire to find firm footing somewhere beyond the bounds of experience' (,^796/8824). It is not easy to see how this inextinguishable desire could be for the falsity of Aj. It is not inconceivable, of course, that we might have a futile yearning for the falsity of an analytic truth. The proposition that 'All men are mortal* may be analytically true, and greatly mourned for all that. So too the proposition that 'There is no triune God', believed and mourned by the new apostate.6 We can always cry for the moon, want the impossible—even the logically impossible. But I say that it is inconceivable that we could have a yearning for the falsity of A3—it is inconceivable that we could have an inextinguishable desire' to consider things abstractly without considering things abstractly. I think this is reason, enough to reject Allison's anodyne interpretation, There is a further problem with the analyticity. If K3 were really just A3, then the question of how anyone or anything could have knowledge of things in themselves would be nonsense. If Allison were right, then Kant would not attempt to say what it could be to have knowledge of things as they are f

" I imagine an apostate unmoved by attempts to save the Trinity by appeal to notions of relative identity, Cf. Peter Geach, c,h, 7 of Logif Matters (Oxford: Blackweil, 1973),

An Old Problem

11

In themselves. He would not attempt to give any content to what is being denied by Kj, But he does. He says, for example, that to have knowledge of a thing in itself would be to be able to ascribe to it 'distinctive and inner predicates' ^5657.8593).7 The fact that Kant says anything at all about this question, is hard to reconcile with Allison's interpretation. On Allison's view, to speak of a thing in itself is ipso facto to speak of a. thing in abstraction from any distinctive predicates whatsoever. Second, there is a problem about causality. Allison says that we render the causal claim about things in themselves innocuous by the simple assertion of A2, which does not commit us to non-spatial, atemporal, unknowable things. That is true. A2 does not require a causal relationship between our minds and some non-spatial, atemporal,, unknowable existents. Allison avoids the traditional problem of affection. But there is a new problem in its place. A2 remains a causal claim: something affects the mind. And the question must be raised whether one is entitled to make any causal claims at all while 'considering things5 at the transcendental level, If Aa is a causal claim, and causality is a category of things considered only in relation to our sensibility, then Allison seems to be failing by his own lights. In asserting Aa, we fail to abstract completely from the relations things have to our sensibility. We fail to keep to the transcendental level, in ascending to the transcendental level, we are supposed to abstract from all the 'conditions of sensibility', namely spacet time, and the categories. But in ascribing causality to things, we fail to abstract from the categories, There is a further problem with the causal claim in A3, Allison's idea, if it worked, would make sense of the claim that things in themselves affect us. But it renders false the Kantian claim that things m themselves are the causes of phenomenal empirical, objects. If a first thing is identical with a second thing, then it cannot be its cause,8 And if the 'something' in A2 is identical with the class of phenomenal objects, referred to in an abstract collective way, then that 'something5 cannot be their cause. I have given grounds for suspicion, but they are not conclusive. They suggest that Allison is trying to see beyond what the philosopher actually said, to what he really meant to say: they suggest that Allison's project, like others before him, is rational reconstruction after all. But Allison does indeed have 7 To have knowledge of a thing in itself would be to know it as 'ein ditrch seine unterscheidenden und inneren Pradikate bestirom bares Ding1 (A 565 76593),a thing determinable through its distinctive and inner, or intrinsic, predicates,

8

This happens to be endorsed by Kant in a very early philosophical work. 'It is inconsistent that anything should have the reason for its existence in itself*, he says in Proposition VI of A New Exposition of (he First Principles of Metaphysical Knowledge (1755), Ak, i, 394, Beck 69, He there draws the implication that God is not his own cause,

12 An Old Problem a solution, where others have failed. And he thinks that non-deflationary attempts are bound to fail. He thinks Kant's problems will go away only if we stop injecting him with the poison of metaphysics. Otherwise we are bound to be left with a philosophy that crudely divides the world into different entities, that supposes an incoherent double affection, that attempts to tell the untellable. If we abandon metaphysical interpretations, and accept Allison's deflationary proposal, we will be saved much philosophical embarrassment. I have suggested that the price of acceptance is too high. If it is, we shall need to find an alternative. I have an alternative. It is not deflationary it does not avoid metaphysics. But it does offer a solution to the old problem.

4. A Metaphysical Proposal, and an Acid Test Passed Are there two worlds, or one world considered two ways? Are appearance and thing in itself the very same? Kant seems ambivalent. Consider the following passage. We call, certain objects, as appearances, sensible entities (phenomena), thereby distinguishing the way that we intuit them from the nature they have in themselves, we place these (considered according to this latter nature). .. in opposition to the former, and .. . call them intelligible entities (noumena). (B3o6)

If we pursue the thread of Kant's anaphors we find one answer. When we call objects, as appearances, sensible entities (phenomena), we distinguish the way that we intuit them from the nature that belongs to them in themselves. The objects that have a nature that belongs to them in themselves are the same as the objects that we intuit. The labels 'noumena' and 'phenomena' refer to the same things. So it seems at first sight. However, one can find in the same passage a different answer. Phenomenal and noumenal 'entities' are described and put *in opposition' to each other, as if they are two nonoverlapping sets of things. One world or two? This ambivalence needs an explanation. The explanation I would like to suggest draws on what Kant says elsewhere, and its defence is the topic of chapters to come. There is one world: there are simply, as Kant says with appropriate vagueness, objects, or things. But there are two, non-overlapping sets of properties. Kant speaks in this passage of the nature that things have in themselves, as he speaks elsewhere of the 'distinctive and inner predicates' of things ^565/6593). The nature things have in themselves is different from what we encounter when we intuit them: the inner or intrinsic predicates are different from the predicates encountered by us. There is one world, one set of things, but two kinds of

An Old Problem 13 properties: Intrinsic properties, and properties that are 'in opposition' to the intrinsic, namely relational properties. The labels 'phenomena' and 'noumena' seem to label different entities, but really they label different classes of properties of the same set of entities. This helps to explain the ambivalence, A distinction between two sets of properties is a metaphysical distinction, but this one has epistemological significance. To have knowledge of a thing in itself, Kant says, would be to be able to ascribe to it distinctive intrinsic predicates ^565/12593), which we cannot do. Ki-Kj should be understood something like this: Mi There exist things in themselves, i.e. things that have intrinsic properties, M.2- The things that have intrinsic properties also have relational properties: causal powers that constitute phenomenal appearances. M.3 We have no knowledge of the intrinsic properties of things. Instead of an inconsistent triad, we have a consistent one. The old and ugly problem disappears. The third claim about knowledge does not undermine the first, two. It has no unwelcome corollaries. Kant's story does not make itself unte.Uable, Kant's existence claim in Ki looked incompatible with the knowledge claim of K3: if we literally have no knowledge of things in themselves, then we do not even know that they exist. If Kj is true, then Ki is false, But interpreted as M.3 and Mi there is no inconsistency. We can know that there are things that have intrinsic properties without knowing what those properties are. Knowledge of things as they are in. themselves involves the ability to ascribe 'distinctive intrinsic predicates' to a thing. That involves more than simply knowing that there are things that have intrinsic properties. Kant's causal claim in K2 looked incompatible with the knowledge claim of Kj. The claim that things in themselves affect us, and that things in themselves are the cause of phenomena, conflicts with the claim that we have no knowledge of things as they are in themselves. If K3 is true, then K2 is false. Interpreted as Mj and M2, however, there is no inconsistency, at least on a certain assumption. On the assumption that causal powers are not intrinsic properties, we can know that a thing has certain causal powers without knowing what its intrinsic properties are. We can know that, things are in certain causal relations with other things without being able to ascribe to them any 'distinctive and intrinsic predicates'. If this is so, there is no problem of affection here. Finally, we do not need to ignore Kant's sense of loss. Kant's attitude to Kj is not the attitude of a man doing conceptual analysis. We have no know-

14 An Old Problem ledge of things in themselves despite having an 'inextinguishable desire' for such knowledge. This attitude makes more sense on the assumption that Kj is really 1X13. There is indeed an entire aspect of the world that remains hidden from us. We are indeed missing out on something. It may be a trivial, analytic thesis to say, with Allison, that we can have no knowledge of things in abstraction from the conditions of knowledge. It is by no means trivial, analytic, that we have 'no insight whatsoever into the intrinsic nature of things- (A277/B333). That is a substantial philosophical discovery, and, in Kant's eyes, a cause for mourning.

2

Three Kantian Theses i. Introduction The problem of the previous chapter, and the two solutions considered, hinge on a distinction that is at the heart of Kant's philosophy, although some have wanted to excise it. Kant says that the 'object as appearance' is to be distinguished from 4the object as it is in itself (B69). It is easy to think that Kant means to draw a distinction between what is dependent on the mind, and what is independent of the mind. Kant's distinction cannot help but conjure up a vision of the veil of appearance, with ideas, or impressions, or representations, on one side of the veil, and things that are independent of us hidden for ever behind it. That vision is at its most provoking in passages like these. External things, namely matter, are.., nothing but mere appearances, that is, representations in us, of whose reality we are directly conscious... Objects... in themselves remain unknown to us ... If I. remove the thinking subject, the whole corporeal world must vanish. (A$j 1—2, A379, AjSj)

Here we have Kant's apparent idealism expressed at its most extreme—and the claim that objects in themselves are unknown to us seems to be part and parcel of that apparent idealism. Kant's philosophy can be made to look like the worst of all veil of appearance philosophies: Berkeley plus unknowable things in themselves. Kant can seem to make the physical world a world of mere ideas, that will vanish on the removal of the thinker. But we know that Kant vehemently rejected the charge that he was a Berkeley in disguise, and he rejected it not simply because the Berkeleyan interpretations ignored the existence of things in themselves, and not simply because they ignored the claims to objectivity established in the Analytic. There is something in principle quite wrong about understanding Kant's distinction in any way that resembles a traditional veil of appearance. That is something I hope to make more evident in time, but for the moment I want to begin by drawing out a different picture, a picture that I introduced in Chapter i, but did not attempt to explain or justify. Let me place a typical expression of Kant's distinction in the company of some passages which I believe illuminate it. (i) This object as appearance is to be distinguished from itself as object in itself. (669)

16

Three Kantian Theses

(2) The understanding, when it calls an. object in a relation mere phenomenon, at the same time forms, apart from that relation, a representation of an object in itsel£ (6306) (3) Concepts of relation presuppose things which are absolutely [i.e. independently] given, and without these are impossible, (.4284/8340) (4) Substance is that which Is ... an absolute subject, the last subject, which does not as predicate presuppose another further subject. (R 5295) (5) The pure concept [of] substance would mean simply a something that can be thought only as a subject, never as a predicate of something else, (A 147/8186) (6) Substances in general must have some intrinsic nature, which is therefore free from all external relations. (A274/B33O) (7) The substantial is the thing in itself and unknown. (jR 5292) (8) We have no insight whatsoever into the intrinsic nature of things. (Aayj/Bjgj) 1 A fuller context for these passages will need to be considered in due course. Each passage, I believe, represents an aspect of Kant's own philosophical view, even when (as in some cases) the context is one in which Kant is discussing the views of others. For now let me merely note that my belief that each of these represents an. aspect of Kant's own view is not particularly idiosyncratic, and Is at least in partial agreement with a number of other commentators.2 1

Note D. P. Dryer's translation of 669; 'The object has then to be distinguished a$ presenting ilselffmm what it is as object in itself (Dryer, Kant's Solution far Verification in Metaphysics (London: George Allen. & Unwin, 1966), 513 n j.) R 5295 and R 5292 (Ak. xviii. 1.45) are dated JVi 1777-80, i.e. probably the period of the writing of the Critiquet whose first edition appeared in 1781. z Compare the following commentators, who seem to agree that Kant is endorsing some of the views he attributes to Leibniz in the Amphiboly of the Concepts of Reflection.; Dryer, who says, 'By a thing in itself is meant any thing whatever considered apart from, relations In which it stands',. KanCs Solution^ 514; Norman Kemp Smith, who says of such passages in the Amphiboly of the Concepts of Reflection, ' K a n t . . . is profoundly convinced of the essential truth of the Leibnizian position'1, Commentary to Kant's Critique f/f Pure Reason (London: Maemillan, 1923), 418-19; Jill Vance Buroker, who says of Aa 84.7 6340, 'It is a conceptual truth about real relations that they presuppose non-relational properties of existing things . . . Leibniz had that much right', Space and Incongruence (Dordrecht: Reidel, 1981), 84 5; Paul Guycr, who says of a passage similar to those quoted (A24Q 52) that 'the very notion of an appearance- -a relation in which something stands to me—implies the independent existence of something which, appears', Kant and (he Claims, 455 n,. 22; Heinz Heimsoeth, who says that Kant denies us 'genuine knowledge of substantiality', in 'Metaphysical Motives in the Development of Critical Idealism*, Moltke Gram, ed., Kant: Disputed Qiteslions (Chicago: Quadrangle Books, 1967); and James van Cleve, who takes seriously, though ultimately rejects, the interpretive possibility of 'things in themselves as things apart from relation*, In 'Incongruent Counterparts and Things in Themselves', G. Funke and T. M, Seebohm, eds., Proceedings of the Sixth International Ka.nl Congress, vol. ii. Pt. 2 {Washington, D.C,: Center for Advanced Research in Phenomenology and University Press of Amer-

Three Kantian Theses 17 The first of the above passages, with Its awkward Idea that we must distinguish an object 'from itself, asserts Kant's fundamental distinction. We must distinguish two aspects of an object: the object as it appears to us, and the object as It is in Itself. We must distinguish the object as It is in relation to us from the same object as it is in itself. The second passage shows how we should understand this. We must, in general, distinguish the way an object is in relation to something else, from the way that object is 'in itself, that is to say, 'apart from that relation'. Kant uses the word 'phenomenon* in this second passage to signify simply 'an object in a relation'. Kemp Smith's translation makes an insertion here: 'an object in a [certain] relation-, implying perhaps that the relation Kant has In mind is exclusively the 'certain' relation that holds between a thing and a human mind. There is no need for this interpolation. Kant has something quite general in mind. There is, however, a prima facie difficulty with this Idea of an object at once existing 'In a relation' and 'apart from that relation'. One natural way of understanding the idea that an object exists 'apart, from a relation' to something else is precisely as a denial of the idea that it bears a relation to something else. This makes Kant's statement incoherent: we must think of an object as at once bearing a relation to something else, and not bearing a relation to something else. We need to find a different way of interpreting the phrase 'apart from that relation'. The third passage gives us what we need, and at the same time supports the very general interpretation we gave to the second passage. Concepts of relation, Kant says, 'presuppose things which are absolutely [i.e. independently] given': that is to say, concepts of relation presuppose the existence of relata whose existence is independent of the relation. That means we must distinguish between a thing as it is in relation to something else, and a thing as it is 'absolutely', or (with Kemp Smith's gloss on schlechthin) 'independently' of that relation. Instead of the incoherent Idea of an object that is at once In a relation to something else, and not In a relation to something else, we have the coherent idea of an object whose existence is independent of its relations to other things. An 'object in itself could exist even in the absence of those other things. It need not coexist with any other distinct thing: Its existence, let us say, does not imply accompaniment. Its existence is compatible with, its being the only thing; its existence is compatible, let us say, with loneliness.^ lea, 1989), 33-45; in 'Inner States and Outer Relations: Kant and the Case for Monadism', Peter H. Hare, ed., Doing Philosophy Historically (Buffalo, N.Y.: Prometheus Books, 1988); and in. 'Putnam, Kant, and Secondary Qualities', Philosophical Papers 34(19(15), 83-109. 3 The terms 'accompaniment' and 'loneliness' are from David Lewis, 'Extrinsic Properties', Philosophical Studies 44 (i 983), 197—200. Lewis uses these to define, not the notion of substance, but Jaegwon Kim's notion of an intrinsic: property, discussed below.

13

Three Kantian Theses

The Idea of a thing that exists 'absolutely' or independently is the idea of a substance. And this brings us to the fourth and fifth, passages: a thing that is an 'absolute subject' is a substance, something that can be thought only as a subject, never as a 'predicate' of something else. Here, as elsewhere, Kant means property by 'predicate*. So the idea of a substance is the idea of something that is an ultimate property bearer, that cannot itself be borne by anything else. The ifth passage says in addition that this is the 'pure concept' of a substance, the concept of a substance in general. The implicit contrast here is with the schematized concept of a substance, about which we shall have more to say in the next chapter. The sixth passage says something more about the pure concept, the notion of a substance in general. 'Substances in general must have some intrinsic nature, which is therefore free from all external relations,'4 Substances,, things that exist independently of relations to other things, 'absolute' subjects which cannot be thought of as predicates of something else, are things that have some 'intrinsic nature* (etmas Jnneres), They must have properties that are 'free' of the substance's external relations {aufiere Verhaltnisse): free of relations the substance bears to things other than itself.5 And they must have properties that are not simply relational, or extrinsic, properties. Substances must have intrinsic properties. What does Kant havein mind here, with this notion of 'something intrinsic" to a substance, something which is 'free of all external relations'? Let us say, as a first approximation, that intrinsic properties are those which do not imply coexistence with any other thing—they are properties which do not imply accompaniment, and (equivalently) are compatible with loneliness, Properties which do imply accompaniment, and (equivalently) are incompatible with loneliness, are the extrinsic, or relational, properties. This understanding of intrinsicness (suggested by Jaegwon Kim, and explored by David Lewis)6 4 1 translate a'uJSere Verhahnisse as 'external relations'. It should convey the Idea of a relation to something else, and should be regarded as neutral as to any doctrine of the internality or externality of relations, construed as a thesis about the reductibility of relations (whether to intrinsic properties of terms, or to terms themselves). The issue of reducibility/irreducibility will be central to the later arguments of Chapters 5 and 6, but because of associations with the British Idealists, I avoid the 'internality/ externality' labels in that dis~ cxtssion. One could translate aufier as outer (as Kemp Smith does), but it is important to avoid a purely spatial implication, 5 A possible contrast here might be a relation, e.g. identity, that a thing bears to itself. (> Jaegwon Kim, 'Psychophysical Supervenience\ Philosophical Sttidits 41 (1982), 51- 70; David Lewis, 'Extrinsic Properties', Lewis, on Kim's behalf, defines accompaniment as the property of coexisting with some wholly distinct contingent object. I find Kim's account of the intrinsic/extrinsic distinction—as interpreted by Lewis-—a close enough approximation to Kant's for present purposes. But as we shall see shortly, if we are to attribute something like it to Kant, the word 'wholly' probably ought to be omitted, to account for Kant's apparent view that spatial properties are extrinsic because part-dependent (and likewise, as we shall

Three Kantian Theses 19 seems like a reasonable first attempt to capture the notion of intrinsicness Kant has in mind. On this way of thinking, there is a close connection between the notion of a substance and the notion of a bearer of intrinsic properties. A substance is a thing which, can exist absolutely, independently of its relations to other things. A substance is the kind of thing that can exist on its own: it can exist and be lonely. But nothing can exist without having properties. If a substance can exist on its own, it must have properties that are compatible with its existing on its own. If a substance can be lonely, it must have properties compatible with loneliness. So a substance must have intrinsic properties. At this stage let us pause and, putting together what we have so far, see what conclusion we reach for Kant's distinction between 'appearance' and 'object in itself*. An object in itself is a thing that exists independently of its relations to other things. An object in itself is a substance, which has intrinsic properties. A phenomenon is an object in a relation to something else. The same object can be described both as phenomenon and as object in itself, precisely because the same object that has relations to other things also has an 'intrinsic nature', if we keep the label 'phenomenon' for the general case of an object in a relation to something else, as suggested in the second of the above passages, then 'appearance' to a human mind can be thought of as a special case. An object that is in a relation to human sensibility is an object that is in a relation: and if we must in general distinguish an object as it is In a relation', from an object as it is 'in itself (6307), then we must also in this case 'distinguish, this object as appearance' from the object 'as object in itself, This understanding of phenomena can be linked with another description offered by Kant in a context where his aim is to distinguish phenomena, not from things in themselves, but from subjective appearances; a context where his purpose is epistemological rather than metaphysical. Phenomena, he says, are appearances that are objective: 'appearances, in so far as they are thought to be objects, in accordance with the unity of the categories, are called phenomena' (A249). Appearances are not phenomena if see in Chapter 4, for Leibniz). Lewis showed the Kim definition to be inadequate: 'lonely* is an extrinsic property, but intrinsic on Kim's definition. However, for a recent and more optimistic defence of a Kim-style definition, see Rae Langton and David Lewis, 'Defining 'Intrinsic", Philosophy and PhenomenologicaI Research, 58 (ic)i>8), Langton and Lewis define a bam intrinsic properly as a property independent of loneliness and accompaniment, where independence means that all four cases are possible; something can have the property and be lonely, something can lack the property and be lonely, something can have the property and be accompanied, something can lack the property and be accompanied. We propose that intrinsic properties supervene on basic intrinsic properties. For an excellent (and different) discussion of Kant's notion of intrinsicness, see van Qeve, 'Putnam, Kant, and Secondary Qualities'.

20 Three Kantian Theses they are not objective, not thought to be objects. The two descriptions are compatible: an objective appearance (a phenomenon according to A249) may also be an object in a relation (a phenomenon according to 6306). Phenomena may be objects in a relation, where the relation is something with which we are acquainted: phenomena may thus be both objective, and appearances, Considered this way, phenomenon may be viewed as the genus of which appearance to human sensibility is the species. Considered this way, Kant's distinction between phenomena and things in themselves is very far from being a phenomenalistic distinction between mental representations and things independent of the mind, it can be summarized thus, Distinction: Things in themselves are substances that have intrinsic properties; phenomena are relational properties of substances, Here we have a metaphysical thesis, and it is the first of the three theses of my title. For ease of reference I shall use it with a capital initial., as a proper name, from now on. Some clarification is called for. If Kant says 'an object in a relation' is to be entitled a phenomenon, and such an object is a substance, then there is good reason for identifying phenomena with substances, rather than with properties. So one could say that Kant's Distinction should instead be understood like this. Things in themselves are substances qua bearers of intrinsic properties; phenomena are those substances qua bearers of relational properties. This, I think, would be basically right. But it would also be potentially misleading, since it would bring with it a temptation to think that substances are somehow in the phenomenal world. There are some delicate issues to negotiate here. This notion of substance is so closely tied to the notion of a bearer of intrinsic properties that Kant can also say, in the seventh passage above, that 'the substantial is the thing in itself' (R 5292, emphasis added). And he can say that phenomena consist 'wholly of relations' ^265/6321). If substance is the substantial, and the substantial is the thing in itself, and the thing in itself is the bearer of intrinsic properties, then any temptation to locate substance—in this sense—in the phenomenal world is to be avoided. For this reason, I propose to identify phenomena with the relational properties of the substances, and with whatever things are constituted by those relational properties. While the first thesis of this chapter is solely metaphysical, the others, while partly metaphysical, also concern epistemology. Consider again the seventh of the passages quoted at the outset, Kant says 'the substantial is the thing in itself and unknown^'(emphasis added). And consider the eighth passage. Kant says that * we have no insight whatsoever into the intrinsic nature of things' (emphasis added), Kant thinks we have no knowledge of things

Three Kantian Theses

21

as they are in themselves. But here that does not express idealism, but something else—the second of the three Kantian theses. Humility', We have no knowledge of the intrinsic properties of substances, These two theses evidently lay behind the argument of Chapter i. Mi and M.2 both drew on the Distinction; the former asserts the existence of things in themselves as things that have intrinsic properties; the latter says that those things also have relational properties that constitute phenomenal appearances. And M3 of that argument amounts to the thesis of Humility. Together, these two theses are what made it possible to avoid that old and ugly problem: to avoid ascribing to Kant tales of the untellable, and to pass what Allison described as the 'acid test* for an interpretation, of Kant's philosophy. This way of understanding Kant helps us to see why he thinks it would be so absurd to deny the existence of things in themselves—and why the wise and sane course adopted by many of his philosophical successors, the course of jettisoning things in themselves, would seem to him so ludicrous, One often inds in Kant an argument that the very notion of an appearance—an objective appearance, a phenomenon-—implies the existence of a thing in itself. Although we cannot know these objects [i.e. objects of experience] as things in themselves, we are required to be able to think of them as things in themselves. Otherwise we should have the absurd conclusion that there could be appearance without: there being something that appears. (Bxxvi) Now we should bear in mind that the concept of appearance . . , itself establishes the objective reality of noumena, and justifies the division of objects into phenomena and noumena, and indeed in such a way that: the distinction does not refer simply to the logical form of our knowledge . , . but to the way in which they are themselves genetically distinct from each other. For if the senses represent to us something merely as it appears, this something must also be a thing in itself. (,^249) It follows naturally from the concept of an appearance in general that something which is not itself appearance must correspond to it. For an appearance can be nothing on its own , , . Unless we are to move in an endless circle, the word 'appearance* must already indicate a relation to something , . , which, however, must be something in itself. (As 51—2) In regarding objects of the senses merely as appearances, we thereby acknowledge that a thing in itself underlies them, although we have no knowledge of how it is in itself . , . Thus, through the very admission of appearances, the intellect acknowledges the existence of things in themselves as well; and we can say that to this extent

22 Three Kantian Theses the thought of such beings underlying the appearances is not only permissible, but unavoidable.7 These passages convey a line of thinking which bears close similarities to what we have just encountered. Notice that they make little sense on a phenomenalistic interpretation, according to which appearances are sense data, or conscious states of mind: what could be easier to suppose than that there could be an appearance without there being something that exists in itself, apart from our states of mind? But if an appearance is a phenomenon, an object in a relation (0307), and if concepts of relation presuppose something that exists in Itself, independently of the relation ^284/6340), then appearance presupposes something that exists in Itself, independently of the relation that is involved in appearance. We have here a particular application of the more general principle. As a general rule, relations, and relational properties, imply the existence of independent bearers; substances capable of existence in the absence of relations to other things, having properties capable of existence in the absence of relations to other things. The inference from appearance to the existence of a thing in itself is an application of this general rule. The Inference is not incompatible with the epistemological thesis of Humility. The final passage, from the Prolegomena^ explicitly distinguishes claims about the existence of things in themselves from, claims about how- the things are in themselves. Although we cannot knowr how a thing is in itself (wie es an sich heschaffen set), we must none the less acknowledge the existence of things in themselves (das Dasein von Dingen an sich selbst). There exist things in themselves, there exist things that have an intrinsic nature. But we do not know them as they are In themselves: we do not know how they are intrinsically constituted,, we do not know what their intrinsic nature is. We can see here what care must be taken when considering Kantian claims about our ignorance or knowledge of 'things in themselves'. Sometimes the phrase (and its German equivalents) may be taken to function as a name for the things: a shorthand for 'things that have an intrinsic nature'. Sometimes it may be taken to work differently, so that 'in themselves5 applies not simply to the things, but to the content of our knowledge of the things: we know them, or do not know them, as they are in themselves. 'We have no knowledge of things in themselves' is then ambiguous between 'we have no knowledge of the things that have an intrinsic nature*, and 'wre have no knowledge of what their intrinsic nature is*. On these disambiguations, the first reading is false: we do have some knowledge of the things that have an intrinsic nature. We know that such things exist. Moreover we are acquainted with 7 Prolegomena, Ak, iv, 314-1.5, Lucas 75-6,

Three Kantian Theses 23 some of the properties of the things that have an intrinsic nature—we are acquainted with their relational properties, which make up phenomenal appearances. And on these disambiguations, the second reading is true; we have no knowledge of what their intrinsic nature is. The issues here are delicate, both philosophically and termmologically: but I shall try to capture the second kind of usage with the phrase 'things as they are in themselves'. It is then no contradiction to say: we have knowledge of things in themselves (I.e. of things that have an intrinsic nature), yet we have no knowledge of things as they are in themselves (i.e. of how they are, intrinsically), Kant seems to believe that our ignorance of things as they are in themselves follows directly from the fact that we have knowledge only in so far as things are given to the senses: 'Properties that belong to things as they are in themselves can never be given to us through, the senses* (.436/652). The epistemological thesis of Humility appears to be linked to a certain view about what the senses can give us. What is at stake, then, in this notion of the senses? To say that knowledge of things depends on the senses is to say that it depends on our being affected by things.8 (Human) intuition ... is always sensible, that i s . . . affected by objects. (435/651) The receptivity of our mind, its power of receiving representations in so far as it is affected in any way, is called 'sensibility', . . Our nature is such that our intuition can never be other than sensible, that is, it contains only the way in which we are affected by objects. (.451/675)

Here we come to the empiricist strand in Kant's thinking, and the third Kantian thesis of this chapter. Receptivity. Human knowledge depends on sensibility, and sensibility is receptive: we can have knowledge of an object only in so far as it affects us, Kant believes that our ignorance of the intrinsic properties of things follows from the fact that knowledge depends on our being affected by things: he thinks that Humility follows from Receptivity. In the remainder of this chapter I shall try to explain and defend these three theses, as interpretations of Kant, and show how they raise a particular problem. Section 2 is devoted to a philosophical exploration of the Distinction, comparing it with a similar metaphysical distinction cautiously 8

The understanding of receptivity developed here conflicts with Wilfrid Sellars's suggestion that receptivity, for Kant, is *a peculiar blend of the passivity of sense and the the spontaneity of the understanding', a suggestion which I do not find very Kantian, See Sellers, Essays on Philosophy and its History (Dordrecht; Reid el, 1974), 49.

24 Three Kantian Theses attributed to Kant by Jonathan Bennett, according to which the thing in itself/phenomenon distinction is, roughly, a substance/property distinction. Section 3 marshals additional textual support for all three Kantian theses, discussing each of the three in turn, and showing how they raise a problem. The problem can. be simply stated: Kant appears to believe that Humility follows directly from Receptivity, but it does not so follow. This is the belief which Strawson calls a fundamental and unargued premise of the Critique? I conclude that the problem to which Strawson draws our attention is one that sorely needs an answer. What is needed, I suggest, is a premise which, conjoined with Kant's Distinction, and the thesis of Receptivity, will imply Humility. What is needed is something that will allow- us to see why, granted a division between the intrinsic and the relational properties of substances, the mere fact that we are receptive creatures implies our ignorance of the intrinsic properties of substances, and thereby implies our ignorance of things as they are in themselves. The search for that missing premise is among the chief tasks of this enquiry, and it is the subject of Chapters 4 to 6, That the notion of a thing in itself functions as the notion of a substance in Kant is not an unfamiliar idea, and has been remarked upon, and complained about, by many commentators, among whom Jonathan Bennett has been particularly enlightening. Something very like the contrast between the Distinction I attribute to Kant, and the phenomenalistie 'veil of appearance* distinction more commonly attributed to Kant, is explicitly drawn in both of his commentaries,10 Although Bennett in the end takes his rational reconstruction down a phenomenalistic path, his discussion provides a useful way of locating the Distinction I want to attribute to Kant, and it also offers a potentially plausible alternative, So the next section has an extended discussion and evaluation of Bennett's suggestion, as a means of gaining a better philosophical understanding of what Kant's Distinction amounts to, with a particular focus on one side of the Distinction: the thing in itself 2. Bennett on Two Distinctions and their Conflation Bennett finds a long tradition of confusion about notions of substance and substratum throughout the history of philosophy, a confusion 'endemic' to the thought of practically all the philosophers he considers, and he finds 9

Strawson, Bounds of Seme, 250. Jonathan Bennett, KanCs Analytic (Cambridge; Cambridge University Press, 1966), 184-7; Bennett, Kanfs Dialectic {Cambridge: Cambridge University Press, 1974), 105-8, The discussion on which I focus is in the latter. The widespread philosophical conflation of the two distinctions is also a theme of his book Locke, Berkeley, Hume (Oxford; Oxford 10

Three Kantian Theses 25 Kant no exception to this confusion. There is a tendency, he says, to conflate the following two distinctions: (a) the distinction between things, and the sensory states which are their appearances; (b) the distinction between substrata, and the properties they support. According to Bennett, the conflation of these two has much to answer for in philosophy's history (Berkeley's idealism being just one of its pernicious effects), and he traces the confusion to two sources. First, a certain slipperiness with such terms as 'idea* in the British tradition enabled the same label to refer both to sensory states and to properties, with the result that substances could be viewed as the supporters of 'ideas' in both of those senses. Second, the apparent similarity of two plausible claims, that we know things only through their appearances, and that we know things only through their properties, paved the way for the identification of appearances with properties. To one in the grip of this confusion, the notion of substance steps in to fill the dual functions of both (a) thing as it is apart from the sensory states which are its appearance, and (b) substratum, or bearer, of properties. Kant belongs firmly within this confused tradition, according to Bennett. He uses 'substratum' sometimes to mean bearer of properties, and sometimes to mean the thing in itself behind the appearance,l J The conflation of properties with appearances is illustrated, he thinks, in these remarks on rational psychology, whose other implications need not concern us here (though we shall return to them in Chapter 10). If I understand by soul a thinking being in itself, the question whether or not. it is the same in kind as matter—matter not being a thing in itself, but merely a species of representations in us—is by its very terms illegitimate. For it is obvious that a thing in itself is of a different nature from the determinations which constitute only its state. (A36o)12

The thing in itself is here contrasted with, on the one hand, representations in us, and, on the other hand, the 'determinations' or properties that 'constitute' the 'state' of the thing itself. Kant's argument appears to move directly from representations in us, to the properties or state of a thing, as if these were the same. He thus appears to conflate Bennett's first distinction (a) above, with the second, (b).

University Press, 1971}, and his article 'Substance, Reality and Primary Qualities', American Philosophical Quarterly* (1965), 1-1:7. 11 Bennett finds examples at 8225, 6231:, A359,6645,6225. 12 This is Kemp Smith's translation; the passage is translated somewhat differently when it is discussed again in Chapter 10,

26

Three Kantian Theses

Bennett is right to say that the conflation of those two distinctions would be a confusion. And he may be right to suggest that the conflation is, if not 'endemic' to philosophy, at least more prevalent than one would like. Postponing for the moment the question, of whether Kant Is guilty, it seems that some of his commentators may be. Heinz Heirnsoeth, more sensitive than many commentators to the metaphysical dimensions of Kant's thought, nevertheless moves indiscriminately between 'appearances' to us, 'representations' in us, and 'properties5 of things in themselves, as if these words and phrases were synonymous,13 This kind of talk is the target of Bennett's complaint. If Kant is guilty of confusion here, he is not the only one. And if Kant is guilty of confusion here, then one might hope that it is the business of his commentators to try to diagnose it and understand it, not merely to perpetuate it. This brings us to the question of whether Kant is in fact guilty as charged. First, suppose that he is. What then? If Bennett is right, then Kant Is interested in two issues which he fails clearly to distinguish. One concerns the gap between sensory states and things independent of them (Bennett's (a) distinction). The other concerns the instantiation of properties in a substance (Bennett's (b) distinction). Kant Is interested in the veil of appearance, and the metaphysics of property instantiation, and fails to distinguish these two issues clearly. The thing in itself, as a substance, is made to fill the two distinct roles of mind-independent reality, and bearer of properties. How should one respond to this (alleged) confusion? Bennett, like many of Kant's critics, advocates ignoring the thing in itself altogether, and therefore ignoring it in either of those two identified roles. We should focus on what Is then left of the first distinction: namely, sensory states. According to Bennett, Kant should be understood m advocating a kind of phenomenalism: objects are logical constructions out of actual and possible sensory states. The (b) distinction, concerning objects and properties, will then have application for these phenomenal objects, but that has nothing to do with the banished thing in itself. That Is not the only possible response. Bennett has discovered, let us suppose, two streams in Kant's thinking. He pursues one, and leaves the other behind. But what of the stream which Bennett identifies and then goes on to ignore? Kant's idea that appearances %rz properties of a substance gets short shrift with Bennett. It will not get such short shrift here, All this Is on the assumption that Bennett's complaint is correct, and that there are in fact two incompatible streams in Kant's thought, But It is far from evident that the complaint is correct. It seems to me that Kant's DisIJ

Heimsoeth, 'Metaphysical Motives', 164, 168, 1.69.

Three Kantian Theses 27 tinction, as I understand it, makes It possible to avoid the mere conflation that Bennett sees. If the thing in itself is a substance, then it will be able to fulfil both roles identified by Bennett. It will be able to fulfil the role of bearer of properties, and the role of mind-independent reality, for if a substance is independent of other things in. general, then, it is, a fortiori, independent of human minds. Moreover, on this interpretation it is easier to move from property talk to appearance talk than Bennett's approach would allow, Bennett assumes that appearances are sensations, or constructions thereof, and on that assumption a move from 'appearances' to properties of substances is a conflation. But we can avoid this problem if appearances are, in the first place, properties of a substance: relational properties of a substance, and in particular, those relational properties with which we can be acquainted. One can interpret Kant's contrast in. A$6o between, the 'nature' of a thing *in itself*, and the 'determinations7 that make up its 4state' or situation (Zus~ land), as precisely a contrast between the intrinsic and the relational properties of the thing. There is nothing in what has been said so far to imply that appearances are, as Bennett thinks, sensations, and there is much to oppose it. This does not, I concede, explain why Kant sometimes wants to move from talk of appearances to talk of 'representations in us\ It does not explain why Kant wants to make colourful couitterfactual statements about the vanishing of the corporeal world. There are possible explanations for these statements, but I am afraid they are to be postponed until the final chapter. I said that Bennett's pair of distinctions resemble the two competing interpretations of Kant with which i began, the traditional 'veil of appearance' view; and the Distinction I attribute to him. That is not quite right. Bennett's first distinction (a) between sensations and mind-independent things does coincide with the traditional veil of appearance distinction. However, his second distinction (b) does not quite match the Distinction that I attribute to Kant. To say that Kant distinguishes a substance from its properties (Bennett's (b) distinction), is not exactly to say that Kant distinguishes a substance as bearer of intrinsic properties from the relational properties of the substance (the Distinction). Bennett's (b) distinction has much in common with the one I propose: both are basically metaphysical distinctions; both provide an alternative to the veil of appearance; both give a role to the thing in itself as substance, and as bearer of properties; both say that phenomena are properties of a substance. But according to the Distinction I attribute to Kant, phenomena are some, not all, of the properties—they are only the relational properties. And things as they are in themselves are not simply substances, but substances as bearers of intrinsic properties. So the Distinction adds to Bennett's (b) distinction a division between different kinds of properties.

23

Three Kantian Theses

Bennett's simpler distinction raises the possibility of a different line of argument for the thing in itself which Bennett himself does not pursue. There are some grounds for attributing it to Kant, and it has one very great virtue, as an interpretation of Kant's distinction: as we shall see in a moment, it leads immediately and uncontentiously to the conclusion that we cannot possibly know a thing in itself, This is a virtue sadly lacking in the Distinction I want to attribute to Kant. Given the apparent similarities between Bennett's substance/property distinction and the Distinction, and this apparent virtue of his, Bennett's substance/property distinction deserves more careful attention. It is worth playing devil's advocate on its behalf, though careful reflection will bring us back to the Distinction in the end.

3, A Case For, and Against, the 'Bare Substratum' Kant may be trying to distinguish substance as property-bearer from the properties a substance bears: that was Bennett's (b) suggestion. On this hypothesis, what Kant means by 'thing in itself is simply substance as property-bearer. If this were correct, then it would place Kant in a long tradition of metaphysicians who hold that such a distinction must be made, Whenever a property is instantiated, it is possessed by a particular or substance that is ontologically distinct from the property it possesses. When we say that the pencil is red, we think of the possessor of the property as being distinct from the redness we attribute to it. But what goes for this property goes for all properties: so we must think of the bearer as distinct from the redness, colouredness, length, shape, smoothness, or any other property we care to name. What then is this substance to which we attribute these various properties? Every candidate answer will 'merely' attribute some further property to the substance. It will not tell us what the substance is. It will 'merely' tell us what properties the substance possesses. This is the 'naked lady' theory of substance, satirized thus by Peter Geach, who says the theory invites metaphysicians to 'mentally strip off the characteristics or properties of a thing to reveal, ultimately, the 'bare particular' in all her glory. But in place of glory there is frustration and bathos: the particular vanishes, leaving nothing at all (except, perhaps, some grist to a feminist mill).14 14 See e.g. Genevieve Lloyd, The.Man of Reason (London; Methuen, :i984)j and Jane Flax, 'Philosophy and the Patriarchal. Unconscious', in. S, Harding and M. Hintikka, eds,, Discovering Reality: Feminist .Perspectives in Epistemology, Metaphysics, Methodology, and Phiimnphy nf Science (Dordrecht: Reidel, 1983), for two among many recent feminist explorations of the political and psychoanalytic implications of metaphor in philosophy.

Three Kantian Theses

29

Geach's conclusion Is that philosophers have, yet again, been beguiled by language: 'some people are so hypnotized by mere words that even without jargon they will think to avoid contradiction by change of the stress: "A substance can have no qualities because it is what has the qualities"!'15 This may be a conception of substance to be found in. Locke as well, explained and possibly satirized by him too in the following famous passage: [I]f any one will examine himself concerning his Notion of pure Substance in general, he will find, he has no other Idea of it at all, but only a supposition of he knows not what support of such Qualities which are capable of producing simple Ideas in us; which Qualities are commonly called Accidents. If anyone should be asked., what is the subject wherein Colour or Weight, inheres, he would have nothing to say, but. the solid extended parts; and if it were demanded, what is it, that the Solidity and Extension inhere in, he would not be in a much better case, than the Indian... who, saying that the World was supported by a great Elephant, was asked what the Elephant rested on; to which his answer was, a great Tortoise; But being again pressed to know what gave support to the broad-back'd Tortoise replied, something, he knew not what. And thus here . . , we talk like Children; who, being questioned, what such a thing is, which they know not, readily give this satisfactory answer, That it is something:, which in truth signifies no more, when so used, either by Children or Men, but that they know not w h a t . . . The Idea then we have, to which we give the general name Substance^ being nothing, but the supposed, but unknown support of those Qualities, we find existing, which we imagine cannot subsist, sine re sub$tante, without something to support them, we call that Support Substantial which, according to the true import of the Word, is in plain English, standing under, or upholding.^

Mockery is a possible response to this metaphysical picture, the response of Geach and (less certainly) Locke, But another response might be to accept the picture, accept the claim that there are indeed substrata, and deny that It is in principle possible to know them. Since knowledge Involves the ascription of properties to a distinct substance, of course the substance in so far as it is distinct from properties will be unknown: there is nothing to know. Questions about what a bare substratum is like do not make any sense, as Bennett says. In short, If the thing in Itself were a bare substratum, then it would be plausible for Kant to both assert its existence and deny knowledge of it. A first thing to be said in favour of this Interpretation is that Heimsoeth, with specialist knowledge of Kant's metaphysics, sometimes seems to

15

P, T. Geach, Truth, Love and Immortality (Hutchinson: London, 1979), 47. A more sympathetic discussion of the bare substratum theory of substance is to be found, in Michael J. Lous, Sub&ttince and Attribute (Dordrecht; Reidel, 1978), 108-12. 1f> Essav It» xxiii, § 3,

30

Three Kantian Theses

attribute it to him. Heimsoeth says that according to Kant 'the finite subject grasps other substances only through , , , "properties1",17 He has in mind such passages as the following: 'Since we know a thing only through its predicates, we cannot know the subject in itself alone' (R 5290).!8 Since this is suggestive of a bare substratum interpretation of the thing in itself, Heimsoeth attributes to Kant the view that we have knowledge only of predicates or properties of things, and therefore do not have knowledge of any substance distinct from those predicates or properties. Unfortunately Heimsoeth at the same time attributes to Kant the view that we are thereby missing out on something. He attributes to Kant the view that there could be such a thing as knowledge of a substance that was not knowledge of its predicates: hence the wtstfulness of tone—'only' through properties'—and the hinted possibility of * genuine knowledge* that is not only through properties. This air of disappointment, which is indeed present in Kant (as I observed in Chapter i), does not sit well with any 'bare substratum1 interpretation of the thing in itself. There is nothing to miss out on.19 Leaving Heimsoeth aside, there may be more to be said in favour of the 'bare substratum' interpretation: there may be textual support for it within the Critique itself. Perhaps it is what Kant means when he says that 'Matter ... is mere external appearance, the substratum of which cannot be known through any predicate . . .' (A359, emphasis added). If Kant really means that the thing in itself cannot be known through any predicate, then this would be perfectly intelligible on a bare substratum interpretation. Moreover, it may be that we can find in Kant precisely the train of thought that Geach describes, particularly if we keep in mind Bennett's hypothesis that Kant may sometimes conflate 'representation' with 'property*. Consider the following typical passage about the transcendental object. If we experimentally substitute 'property1 for 'representation' (Forstellung)y taking seriously Bennett's hypothesis, we have a line of thinking which appears to have close affinities with the bare substratum view.

17

^Metaphysical Motives', 164, my emphasis. Ak. xviii. 146, M 1777-80, 19 There are some reasons for thinking that this is not. Heimsoeth*s considered, or at least not his only, interpretation. For Heimsoeth appears to run together a number of different interpretations of the thing in. itself, and while it is possible (though uncharitable) to believe that Kant holds them all at once, Heimsoeth does not seem alert to their differences. In addition to the 'bare substratum' idea of the thing in itself are two other very different ideas, one of the thing in itself as having intrinsic or *in.ner properties' of which, we are necessarily ignorant, and an.oth.er idea of the thing in itself as having essential properties of which we are necessarily ignorant. All three may have interpretive merits, but. they are hardly the same. 18

Three Kantian Theses

31

All [properties] are, it is true, referred by the understanding to some object; and since appearances are nothing but [properties], the understanding refers them to a something, as the object. . . But this something, thus conceived, is only the transcendental object; and by that is meant a something = x, of which we k n o w , , , nothing whatsoever, but which ... can serve only for the unity of the manifold . . , By means of this unity the understanding combines the manifold into the concept of an object. This transcendental object cannot be separated from the , . . data, for nothing is then left through which it might be thought. Consequently it is not in itself an object of knowledge. (Aago)

On this interpretation, a transcendental object, em Etwas = >v, is the particular to which properties are ascribed, the item to which the manifold and various properties are attributed, the 'something' conceived of as bearer of properties, but not really separable from the properties we attribute to it. Independently of the properties we attribute to it, we can think of it only as a 'something': not an object of knowledge. It is tempting here to borrow Geach's parody; the transcendental object has no properties because it is what has the properties. But that would be unfair to Kant, I think. He would be guilty of no contradiction if he thought of the transcendental object as bare substratum, as far as I can see. The transcendental object would be knowable in so far as we attribute properties to it, but conceived of as a particular 'something' distinct from the properties, not knowable. Can. we identify Kant's transcendental object with the thing in itself? Kant's doctrine of the transcendental object has been frowned upon. Kemp Smith, for example, dismisses it as a 'pre-Critica! or semi-Critical survival [which] must not be taken as forming part of Kant's final and considered position'.20 Whether or not that is so, what seems clear, and is indeed emphasized by Kemp Smith, is that Kant does use "transcendental object' and 'thing in itself as synonyms.21 So passages like A 250 might well be thought to support the interpretation we are considering: the transcendental object, and therefore the thing in itself, is a bare substratum. Finally, it is worth noting in its favour that the 'bare substratum' interpretation may, at a stretch, be compatible with the first three of the four passages with which I began this chapter. This object as appearance is to be distinguished from itself as object in itself. (B6g) The understanding, when it entitles an object in a relation mere phenomenon, at the same time forms, apart from that relation, a representation of an object in itself. (B3o?) m

Kemp Smith, Commentary t 218. See, for example, A191 /JB2.J6, A277/B333 ff., ,^.288/6344, A35&, ,4361, Aj66, A372, ,4379, Ajijo, Aj93, A 3(14. These are all cited by Kemp Smith. 21

32

Three Kantian Theses

Concepts of relation presuppose things which are absolutely [i.e. independently] given, and without these are impossible. (,^.284/6340)

The idea of an object in itself could conceivably be interpreted as the idea of a property bearer that is distinct from (exists apart from, independently of) any properties whatsoever. It could therefore conceivably be interpreted as the idea of a 'bare substratum', However, the bare substratum interpretation creaks and strains at this point, and it collapses completely when we include the fourth passage. Substances in general must have some intrinsic nature, which is therefore free from ail external relations. (.4274/6330) If the thing in itself is a substance, then it is not a substance conceived of as a bare substratum, but a substance conceived of as having 'some intrinsic nature', etwas Inneres, And if we stop cheating on behalf of the devil and fill in the ellipses in the passages at A359 and A.250, we find that "matter is mere external appearance, the substratum of which cannot be known through any predicate that me can assign to if (A359, emphasis added). Kant does not mean that the substratum cannot be known through any predicate at all. He means that it cannot be known through any predicate or property that we are in a position to ascribe to it. And when we turn to Azso, we find that Kant means by 'transcendental object*, 4a something = x, of which we know, and with the present constitution of our understanding can know, nothing whatsoever ...' (A.25O, emphasis added). It is quite evident that Kant thinks that there is something to know, but that we cannot know it—not at any rate with the 'present constitution of our understanding'. This is incompatible with the bare substratum interpretation. And the passage from the Refiexionsn can be reinterpreted in light of this as an expression of Humility: 'since we know a thing only through its [relational] predicates, we cannot know the subject as it is in itself alone' (R 5290). In short, the point made against Allison in the last chapter holds equally here. It is patently obvious that Kant thinks we are missing out on something: there is something to know that we cannot know. That; is why he is a chastened and disappointed metaphysician, as Heimsoeth rightly implied, but wrongly explained. For it makes no sense to suppose that anything could in principle—no matter what the 'constitution' of its understanding (A25o)—have knowledge of a bare substratum. My conclusion is that, while the 'bare substratum' interpretation of the thing in itself has some prima facie plausibility, it fails. The thing in itself is not unknowable in virtue of its 'bareness* of properties: it is unknowable in virtue of its having intrinsic properties that we cannot know.

Three Kantian Theses

33

4. Three Theses: Some Further Tex is So far I have described Kant's Distinction rather abstractly, with little attention to the question of what phenomena and things in themselves are like,, which now emerges as the question of what the relational and intrinsic properties of substances are. The task of remedying this will be a long one, and will take us beyond the scope of this chapter. We can begin, though, by marshalling some additional texts, some of which are more complete versions of those considered earlier. Our interest is in their implications for all three theses, to be taken one at a time, (9) The Intrinsic and Extrinsi£.-—In an object of the pure understanding the intrinsic is only that which has no relation whatsoever (so far as its existence is concerned) to anything different from itself. It is quite otherwise with a substantial phaenomenon in space; its intrinsic properties are nothing but relations, and it itself is entirely made up of mere relations. We are acquainted with substance in space only through forces which are active in this and that space, either drawing other objects (attraction) or preventing their penetration (repulsion and impenetrability). We are not acquainted with any other properties constituting the concept of the substance which appears in space and which we call matter. As object of pure understanding, on the other hand, every substance must have intrinsic properties and powers which concern its inner reality. (A26s/B32i) (10) Substances in general must have some intrinsic nature, which is therefore free from all external relations ... But what is intrinsic to the state of a substance cannot consist in place, shape, contact, or motion (these determinations being all external relations). 0^274/13330) ( I T ) Matter issufota-ntwphaenomenon. I search for that which belongs to it intrinsically in all parts of the space which it occupies, and in all the actions it performs, and these of course can only be appearances to external sense. So I have nothing that is absolutely intrinsic, but only what is comparatively intrinsic, and that is itself again constituted by external relations. It is silly to suppose that matter has an absolute intrinsic nature of the sort conceived by pure understanding, for matter is not an object of pure understanding. On the other hand, the transcendental object which may be the ground of this appearance that we call matter is a mere something. We would not understand what it was, even if someone could tell us ... If the complaints that Sve have no insight whatsoever into the intrinsic nature of things' are supposed to mean that we cannot grasp by pure understanding what the things which appear to us may be in themselves, they are completely unreasonable and stupid. They want us to be able to be acquainted with things without senses! (Aayj/Bjjj) (12) All that we know in matter is merely relations (what we call its intrinsic prop-

34

Three Kantian Theses erties are intrinsic only in a comparative sense), but among these relations some are, . . enduring, and through these we are given a determinate object... It is certainly startling to hear that a thing is to be taken as consisting wholly of relations. Such a thing is, however, mere appearance, (A.285 76341)

(13) Everything in our knowledge which belongs to intuition . . . contains nothing but mere relations, of locations in an intuition (extension), of change of location (motion), and of laws according to which this change is determined (moving forces). What presents itself in this or that location, or, beyond this change of location, what activities occur within the things themselves, is not given through these relations. Now through mere relations one cannot be acquainted with a thing as it is in itself. We may therefore conclude that since external sense gives us nothing but representations of mere relations, this sense can contain in its representation only the relation of an object upon the subject, and not the intrinsic properties that belong to the object as it is in itself. (E6j)'2

These passages are taken from Kant's discussion of metaphysics in The Amphiboly of the Concepts of Reflection, except for the last, which is from the Transcendental Aesthetic. 5. The Distinction

Consider the subtitle in (9), 'The Intrinsic and Extrinsic* (Das Inmre uml AujSere}. Kant draws a distinction between two classes of properties (Eigensfhaflen, Bestimmungen, Akzidenzen). lie draws a contrast between properties a thing has when it has 'no relation whatsoever... to anything different from itself, and properties a thing has when it does have some relation to something 'different from itself. He thus draws a distinction between properties that are intrinsic and those that are not. Kant does not distinguish very clearly between relations and relational properties, referring to forces, for example, as both. But a basic distinction is clearly being made between the intrinsic and extrinsic, or relational—somehow construed. We shall have more to sav about this in a moment. The distinction itself is not offered bv 22 The fourth sentence in passage (u) is more literally: 'the absolute intrinsic nature, according to pure reason, of matter, is a silly idea, a mere fantasy (Grii!e)\ With respect, to the first ellipses of passage (12), Kant actually says 'selbstandige und beharriiche*, which Kemp Smith translates 'seif-subsistent and permanent'. Given that the purpose of the passage is to show among other things that phenomenal substance is precisely not selfsubsistent in the manner of a true substance, I would translate this as 'independent and permanent1, where the independence in. question is not taken to be absolute independence, but independence from the flux, of change, in line with the First Analogy and the B Refutation of Idealism, Independence of this kind is required for the arguments of the Analytic, but the absolute independence of a thing 'that exists apart from ail relation* is not. This point is developed in the argument, of the next chapter.

Three Kantian Theses 35 Kant as something novel. As a matter of fact, it Is present in the metaphysics textbook that Kant himself used, in teaching, for most of his life. Baumgarten distinguishes intrinsic properties from extrinsic, describing the former as inner, absolute determinations, the latter as external, relative determinations, or (equivalently) relations,23 While the distinction between two classes of properties is not offered as something novel, something else is. What is novel—what is indeed startling^ according to Kant—is his own proposal that there is something that consists wholly of relations. Matter^ phenomenal substance, 'consists wholly of relations' (12), is 'constituted by external relations* (i i), is 'entirely made up of mere relations' (9), is *nothing but mere relations' (13). He refers briefly in (9) to the 'intrinsic properties' of matter, but that is swiftly undermined: what we call its intrinsic properties are intrinsic only in a comparative sense, and it is silly to suppose that matter is the sort, of thing that could have an absolutely intrinsic nature (r i and 12). In all of these passages there is a clear expression of one side of the Distinction: phenomena are constituted by relations, or relational properties,24 What of the other side of the Distinction? Consider (13), an important and complex passage to which we shall have reason to return. What presents itself as matter in space is not given through the relations that constitute matter: the intrinsic properties that belong to the object as it in itself are absent from matter in space. Any activities occurring within the things themselves are not given through, the relations, Kant seems clearly to be talking about two aspects of the one thing: the thing that presents itself to us as matter—as something constituted by relations-—is the same as the thing that has 'activities1 occurring within it, 'intrinsic properties belonging to the object as it is in itself. In (i r) this thought recurs. There is a transcendental object, a ground of this appearance that we call, matter, a ground which has an intrinsic nature; but the object is for us a mere something, since the 23

Baumgarten says that relations are external, 'determinations', which may be why, in Kant's usage, relations seem to be viewed as properties. *§ 37. Rdationes [Verhaltnisse], . , sunt . . , determinaii&nes exiemae [aussre] (reiativae, ad extra? extrinsecae), reliquae omnes, internae [inure Bestim.mun.genJ.' A. G. Baumgarten, Metaphysica (1757), reprinted in the Academy edition of Kant's works, Ak. svii. 35. Square brackets contain the original German annotations, Baumgarten also says that if determinations are considered as belonging to a thing in the absence of a nexus they are 'absolute', whereas if they are considered to belong to a thing only when it is in a nexus, they are 'respective*. Ermanno Bencivenga draws attention to Baurngarten's text in Kant's Copernuan Revolution {New York: Oxford University Press, 1981), 32, 24 For a different interpretation, of these passages see Harold Langsam, who argues that properties of phenomenal things are intrinsic, but relational in their concepts, in 'Kant, Hume, and Our Ordinary Concept of Causation*, Philosophy and Phenomena logic a I Research 54(1994), 625-47.

36

Three Kantian Theses

Intrinsic nature is unknown. Passage (9), In light of these, bears a similar interpretation. The substance which appears in space and which we call matter must also have another aspect; it must have intrinsic properties and powers that concern Its inner reality. These passages have implications for different aspects of our understanding of phenomena: for phenomenena as external appearance; for phenomena as matter; and for phenomenal substance. The irst two of these topics can be attended to here; the third is a large one, and needs the more detailed attention I give it in the next chapter. Kant often refers to phenomena as external appearance (dnflere Erscheinung\ and one might wonder whether Kant's discussion in the passages above sheds any light on this notion. The phrase is usually interpreted in a spatial sense-—'outer appearance*. And the phrase is usually interpreted in such a way that 'external' means 'external to ui. The notion of an external appearance is thus taken to convey the idea of appearances outside us, in space, However., Kant's discussion in the above passages gives us reason to reconsider the notion of external appearance, and to ask whether it may sometimes mean something rather different. Kant here describes a certain class of properties as extrinsic (auSer, or nominalized as das Au/ere). If appearances are external because they are phenomena, and phenomena are external because they are extrinsic, or relational, properties, then the usage of au/er in the notion of 'external appearance* may well be connected with the usage of aufier here. There are two potential ambiguities, The first is noted by Kant himself elsewhere, who says that the idea can sometimes carry an implication of outerness and sometimes an implication of otherness (A373). Suppose we deem 'external' to be an equally ambiguous English translation: then to say that x is external toy can be to say that x is outside j, located in a different part of space to j; or it can be to say that x involves something other than j, something distinct from y. These two notions are not the same, though it is part of the argument of Kant's Aesthetic that the two are closely connected in human experience,25 Kant says: 'By means of outer sense . .. we represent to ourselves objects as distinct from us (aufier urn), and all without exception in space' (A22/B37), Kemp Smith's translation of the aufier tins is spatial, 'outside us', but this makes a tautology of something Kant does not appear to be offering as a tautology,26 What Kant means by outer sense, or external 25 This connection has been fruitfully explored by many, including Strawson, .Bounds of Sense", Strawson, Individuals (London; Methuen, 1959); Gareth Evans, 'Things Without the Mind', in Zak van Straaten, ed., Philosophical Subjects (Oxford; Oxford University Press, 1980), 76-116; Bennett., Kant's Analytic; Guyer, Ka-nl and lite Claim.1;.. 2(J O*. Guyer, Kant and the Claims, 3 i i.

Three Kantian Theses 37 sense, is our capacity to be affected by things external to us, distinct from us, other than us. By means of outer, or external, sense, we represent these distinct objects that affect us as distinct from us, by representing them in space. The Auflere of the subtitle is potentially ambiguous in a similar manner,. between having to do with space, and having to do -with, something eke, 'External' can have a spatial sense, or it can have the more general sense of extrinsic (when Kant speaks of 'relational or external properties'-—Verhaltnis- oder aufleren Bestimmungen (63397 A283)). It is important to avoid the mistake of taking it for granted that extrinsic properties are ipso facto spatial properties. And when Kant speaks of external appearance throughout the Critique, it should not be taken for granted that by 'external' he means simply 'in space'. There is a second ambiguity to note. Assuming that * external' is sometimes to be taken as meaning 'extrinsic*, a further question is: external to what? In A22/B37 , Kant clearly means external to us—but in the numbered passages listed above 'external' means external, or extrinsic, to a substance. To be sure, the properties in question of shape, contact, attraction, repulsion, are also external to us: but what is relevant, given the contrast Kant wants to draw, is that they are external, or extrinsic, to a substance, a thing in itself, whose intrinsic properties are unknown. When Kant refers throughout the Critique to external or outer appearance (auftere Erscheinung), it is not always evident which of the two possibilities he means. But if appearances are constituted by properties of a substance, and in particular, by extrinsic or relational properties of a substance, then while 'external appearance' may well be external to us, what Kant may sometimes mean is that appearance is external to the substance, the thing in itself. This understanding of Kant thus allows us a rather different interpretation of the notion of an external appearance. To say that we know the external appearances of things, but not the things as they are in themselves, is to say that we know the extrinsic, but not the intrinsic, properties of substances. There are also implications here for our understanding of the realm of phenomena as the realm of matter. Phenomenal properties are physical properties, and Kant says in these passages that physical properties are all relational. Kant mentions two kinds of physical property, in (9) and (13): spatial properties of 'place, shape , , , motion', and dynamical properties, forces of attraction and repulsion, or impenetrability. These, he says, are 'all relations1, 'nothing but relations', 'mere relations'. Elsewhere he says that what is given to us in space are 'merely relations, formal, or also real' ^284/6340), again distinguishing between spatial (formal) and dynamical (real) relations. In what sense are spatial and dynamical properties 'nothing but relations'? Kant seems to believe that spatial properties, even apparently intrinsic

38

Three Kantian Theses

properties such as shape, are extrinsic, or relational, because they depend on the parts of the shaped body: 'Corporeal things are never anything but relations only, at least of their parts external to one another' ^283/13339). This suggests that Kant's understanding of intrinsicness is to be interpreted in a particular way. It was suggested that an intrinsic property is a property something can have even when unaccompanied by another distinct thing: but what counts as another distinct thing? A proper part of a thing is not identical with the whole it is part of: so it is, in one sense, a distinct thing, even if not a wholly distinct thing. If the spatial properties of an extended object depend on the parts of the object, and if a part is a distinct thing for the purposes of Kant's notion of intrinsicness, then spatial properties are extrinsic properties. So perhaps that is how Kant*s notion of intrinsicness is to be understood, Dynamical properties are relational for very different reasons, in explaining why they are 'mere relations', Kant says that forces 'are active in this or that space, either drawing other objects (attraction) or preventing their penetration (repulsion and impenetrability)* ^265/6321). Now this notion of force (Kraft) which is so central to Kant's matter theory is not easy to classify; or rather, there seems to be more than one feature of the physical world that is picked out by his notion of force. Is force a relation—for example, the relation a thing bears to another when a thing attracts another? Or is it a causal power to enter into such relations—for example, the power a thinghas to attract another? Kant seems to think of forces as both. He thinks of them as relations (of attraction and repulsion), and as causal powers (of attraction and repulsion)—and either way, as relational properties, Certainly the notion of force, however construed, is a causal notion. Kant describes it as a predicable of the relational categories of causality and (in the case of reciprocal attraction and repulsion) community (A82/Bio8). Some philosophers in our own time have argued that force is to be understood as a species of the causal relation itself.27 The notion of force can also be understood as the notion of a power—which is itself to be thought of as a certain kind of relational property. Forces, construed this way, are relational, because they are powers to act in certain ways towards other things: something has a force of attraction when it has a power to attract something else; something has a force of impenetrability when it has a power to repel something else. Something can have a power to attract without actually attracting, so the power is not to be identified with the relation itself. All the same, there is something relational, something other-directed, about the concept of a force, construed as a power. 27

See, for example, John Bigelow, Brian Ellis, and Robert Pargetter, 'Forces*, Philosophy of Science 55 (1988), 614—30.

Three Kantian Theses 39 There is, we can say, something conceptually relational about a force, thus construed: something has an attractive power if and only if it would attract other things in certain circumstances. That is a fact about the concept of a force, or power. Does that prove that force, construed this way, is an extrinsic property? No, The relationaJity of concepts is not the same as the extrinstcness of properties. Whether a power is to be classified as an extrinsic property will depend on what we take intrmsicness to be: and on the definition so far considered, it will depend on whether something could have a power in the absence of other distinct things. Answers to that complex question may well differ, even among philosophers today—so I propose for now to postpone considering it until a more appropriate occasion arises (in Chapter 5). What seems clear enough, and adequate for our present purposes, is that forces*-—understood as causal powers-—are regarded by Kant himself as properties that are not only conceptually relational, but also extrinsic.28 We here encounter Kant's dynamical theory of matter, which he endorsed in some form throughout the length of his philosophical career, and which forms a backdrop to some of the more puzzling parts of the Critique. This theory, and some of its merits, will be discussed in more detail later (Chapters 5 and 8). According to Kant, matter is constituted by fundamental forces of attraction and repulsion. Matter requires the interaction of both forces: for if attraction existed alone, matter would coalesce to a point, and if repulsion existed alone it would be dispersed to infinity. Matter is thus constituted by a 'conflict' of forces: Since all given matter must fill its space with a determinate degree of repulsive force, in order to constitute a determinate material thing, only an original attraction in conflict with an original repulsion can make possible a determinate degree of the filling of space, which is matter.29

Kant sees matter as made up, not of atoms, but centres of force. Since forces are relational properties, according to Kant, and since the realm of matter is the phenomenal realm, the phenomenal realm is thus constituted by relational properties, just as the Distinction says. And this brings us back to the previous point: the realm of matter is what Kant describes as 'external 28

Lloyd Humberstone advises that the 'intrinsic/extrinsic* distinction be distinguished from the 'non-relational/relational' distinction, the former being a label for properties of properties, the latter for properties of concepts. However, I use 'relational* and 'extrinsic' as synonyms, for convenience, and Humberstone's label of 'relational' becomes my label of 'conceptually relational'. The distinction becomes especially important for cases where intrinsic properties are designated by relational concepts: e.g. the property of being shaped like the Eiffel Tower, (Some philosophers may think powers may belong in this category, but Kant, so I argue, does not.) See Humberstone, 'Intrinsic/Extrinsic', Synffiese 108 (iyy6), 205-67. 2 250,

Three Kantian Theses 45 thesis looks easy. What could it mean for something to be 'given to us through the senses1 (A.^6/B^2) if not that we are receptive to and affected by objects? The thesis of Receptivity,, its abstraetness notwithstanding, looks obvious, trivial. It is not, though. One can easily imagine a competing empiricist picture, a cautions phenomenalism perhaps, that turns its back on metaphysicalsounding talk of human faculties (whether active or passive), admits the existence of sense data, professes agnosticism about causality generally, and therefore questions the meani.ngfulness of statements asserting causal relations between sense data and the world. Such a phenomenalism32 would admit that something is given to the senses, but admits no passive sensibility. It would admit that something is given to the senses without thereby admitting that something affects, the senses. It does not go without saying that an empiricist presumption must express itself in a carnal thesis like that of Receptivity, Kant's endorsement of Receptivity, taken on its own, would yield a particular kind of empiricism: something that more closely resembles a causal theory of knowledge than a phenomenalism, Kant does not think it goes without saying that knowledge requires a passive sensibility either, and he writes of a different contrast, one that has already been implicitly encountered in the contrast between the sensible and the intelligible. When Kant asserts of human beings that we have a passive sensibility, he compares that with another way of achieving knowledge, impossible for us, but not. for God, who has an 'original' or 'active* intuition, and is a being to whose view 'all things lie open . , , with total clarity*.33 Intelligible entities are indeed intelligible—to him. Our way of intuiting is dependent upon the existence of the object, So it is possible only if the subject's faculty of representation is affected by thctl object,,, However universal this may heT it does not: cease to he sensibility. It is derivative (iniuitus derivalivus), not original (inluitus originariw\ and therefore not an intellectual intuition . ., Such intellectual intuition . , . can never he ascribed to a dependent being. (672, first italics added)

Here, in the concept of the original intuition, we strike depths which I have no ambition to plumb, merely pausing to note the significance which Kant evidently attaches to his apparently obvious premise of Receptivity. A further sign is to be found in the startling implications he takes Receptivity to have for knowledge of the mind itself. 32

Perhaps Hume, perhaps the Russell of the logical atomism phase ('Philosophy of Logical Atomism' (1918), in R, G Marsh, ed., L&gic and Knowledge (London: Allen & Umvin, 1956)1 175-281). Most relevantly for Kant (as we shall sec in the ensuing chapters) it Is Leibniz who denies Receptivity. 3S New Exposition (1755)5 Ak, i, 391, Beck 64, an early expression, admittedly.

46 Three Kantian Theses If the capacity of becoming self-conscious is the capacity of seeking out (apprehending) what lies in the mind, the mind must affect itself... it then, intuits itself.,. as it is affected by itself, therefore as it appears to itself, not as it is. (669) Here again we strike deep and murky waters, and here again my aim is simply to emphasize the seriousness with which Kant treats his apparently obvious premise of Receptivity, and to emphasize the very general implications he sees it as having for Humility in every sphere of knowledge. The most fundamental fact about our intuition is that it is receptive: our way of intuiting is dependent upon the existence of the object, and so it is 'possible only if the subject's faculty of representation is affected by that object'. Kant says that 'our intuition can never be other than sensible, that is, it contains only the way in which we are affected by objects' (A,5i /By5), If this is indeed the most fundamental fact about our intuition., then when, Kant says that an existing thing is presented in intuition, he means, first and foremost, that it affects our sensibility. We only know objects in so far as they affect us; therefore we do not know them as they are in themselves. We have knowledge of things only through the senses: therefore we do not know their intrinsic properties. Kant seems to think these are valid inferences, but if Straw-son is right, he gives us no good reason at ail for thinking they are. In the absence of any justification from Kant, it might be possible to safely disentangle the good Kantian empiricism from its dubious alter ego, which is just what Strawson does. His Kant is a kind of Janus, sane and mad, whose benign empiricist face vies with the face of an ugly idealism, which invokes a supersensible realm only to banish us from it for ever. Strawson thinks one face can be safely ignored. Kant thinks there are not two faces but one. He thinks that because we have knowledge only through the senses, there must be aspects of reality that are necessarily beyond our grasp. He thinks that Receptivity implies Humility. If Kant is right, we cannot separate these two in the way that Strawson hoped, At the other end of the interpretive spectrum we find, again, Heimsoeth, who embraces the metaphysical Kant with open arms. He shows some testiness towards the empiricist thieves who have stolen Kant in the misguided belief that he is one of their own. 'One can no longer believe', he says, 'that Kant's "critical" attitude can be taken as a model for one's own avoidance of all metaphysical problems,'34 Heimsoeth, with formidable scholarship, traces every wrinkle on the face that so repelled Strawson, and in the process gives full acknowledgement to Kant's inference from Receptivity to Humility. But he acknowledges it and simply endorses it. He offers this as his own 34

'Metaphysical Motives', 158,

Three Kantian Theses 47 commentary on Kant's inference: 'finite knowledge , .. must make contact with, its object only by Receptivity, It follows from this that no finite knowledge is knowledge of the thing as it is in itself.'35 This is simply to assert what Strawson called the fundamental unarguecl premise. Kant certainly believes that 'it fellows from this*: that is to say, Kant believes our ignorance of things in themselves follows from the fact of Receptivity, But we want to know why. Heimsoeth gives no answer, but repeats Kant's inference without philosophical qualm. One might wish he had taken his own severe advice: none of Kant's attitudes, whether critical or otherwise, can be taken as excuses for one's own avoidance of metaphysical problems. To take Kant's more enigmatic statements and repeat them in assertoric mode does little to advance understanding. Humility does not, contrary to Heimsoeth, simply follow' from Receptivity. Strawson is right to say that Kant's inference, as it stands, presents a problem. Humility does not follow from Receptivity: or not, at any rate, without some other premise. But what? This brings us to new business. What is needed, I suggest, is a premise that will allow us to move from the premises we already have to the conclusion about Humility, On the assumption that there is a distinction between the intrinsic and relational properties of things, and on the assumption that human, beings are receptive creatures, we need something to fill, the gap: something that will show us why, from this starting-point, we have no knowledge of the intrinsic properties of things. Before embarking on this journey, however, there is a potential challenge to meet. For the texts we have just been considering make use of a notion of wbstantia- phaenomenon^ which has some as yet unremarked implications. If the contrast between things in themselves and phenomena is supposed, in part, to be a contrast between substance and not-substan.ce, what is to be made of this notion of phenomenal substance? 35

Ibid. 162.

3 Substance and Phenomenal Substance I. Introduction Does the Distinction I attribute to Kant conflict with Kant's commitment to phenomenal substance? In this chapter I shall address that question by exploring the status of substance as phenomenon. This is a large topic, and my aim is not to analyse or evaluate the arguments of the First Analogy, nor to begin to do justice to the vast and complex literature on this subject, but rather to show how Kant's Distinction is compatible with a commitment to phenomenal substance as something enduring in the appearances. In the process of addressing this problem, more support is offered for two theses of the previous chapter: the Distinction itself, and the thesis of Humility. Some cautionary remarks are in order. The notion of a 'phenomenon' tends to call to mind nowadays the notion of something of which we are aware. Because of its historical association with philosophical doctrines that take the name "phenomenalism7 it calls to mind sense data, and logical constructions thereof. Because of its association in the philosophy of science with the phenomena or observable regularities that may or may not need to be saved, it calls to mind what can be seen or at least detected. However it has not always been so, and it seems to me that in Kant's own rationalist philosophical background the primary sense of phenomenon is metaphysical: to say that something is a phenomenon is to say that it is not fundamental, that it is in some way dependent or derivative. The Distinction offers a metaphysical conception of phenomena. Kant says, in a quite general manner, that the understanding tends to 'entitle an object in a relation mere phenomenon* (6307); and confirmation of a broadly metaphysical conception of phenomena will be found in this chapter. 2. The Pure Concept of Substance vs. the Schematized To attribute to Kant the view that things in themselves are substances is to ignore the conception of substance for which Kant is most famous, and for which he argues in the First Analogy. The Distinction draws on a general conception of substance which is also to be found in Kant's writings, which is not the same as the substance of the First Analogy. This general conception, or pure concept., of substance is described in such passages as these:

Substance and Phenomenal Substance

49

The pure concept [of] substance would mean a something that can be thought only as a subject, never as a predicate of something else. (Ai47/Bi86) Substance is that which is . . . an absolute subject, the last subject, which does not as predicate presuppose another further subject. (R 5295^ Substances in general must have some intrinsic nature, which is therefore free from all external relations. (.4274/13330)

Kant says, at the end of the chapter on Schematism, that we can find no use for this pure concept of substance in experience. The concept of substance can be applied to the empirical world only when it is interpreted through the schematism: The schema of substance is permanence of the real in time, that is, the representation of the real as a substratum of empirical determination, of time in general, and so as enduring when all else changes. (Ai43/Bi83)

We apply the concept of substance in experience when we represent 'the real* that is given to us in experience as a substratum, or bearer, of properties that change over time, and as something that endures throughout all this change. Substance, once the sensible property of permanence is taken away, would mean simply a something that can be thought only as a subject, never as a predicate of something else. Now I cannot put this representation to any use, because it doesn't show me at all which properties belong to the thing whose role is to be a first subject of this kind. (Ai47/Bi86)

Kant says that we cannot put the pure concept of substance to use in a way that will allow us to have knowledge of a substance of this kind. The reason is that we are not shown 'which properties belong to the thing whose role is to be a first: subject'. What properties are we not shown? Evidently, the properties required in something 'whose role is to be a first subject'. And what properties are required in a first subject, a substance that conforms to the pure concept? The answer is clear: intrinsic properties. A substance that conforms to the pure concept is an independent thing, that can exist by itself, and must thus have properties compatible with its existing by itself. We are not able to put the pure concept of substance to use in a manner that will yield knowledge, because we are not shown the intrinsic properties of such a substance. This understanding of the above passage is in keeping with Kant's statement about what would be involved in having knowledge of a thing as it is in itself. Such knowledge, he says, would involve being able to 'think of it as a thing that can be determined through its distinctive and 1

Ak, xviii, 1:45, M 1777-80,

50

Substance and Phenomenal Substance

Intrinsic predicates' (^565/6593). To 'determine' a thing is to ascribe predicates or properties to it,2 and such knowledge would involve determining the substance by ascribing to it distinctive intrinsic predicates. Kant says that we can determine objects of the senses, but only with, respect to the predicates that are possible in. the field of appearance-—which he says are all relational. In the absence of acquaintance with intrinsic properties, we are unable to determine things as they are in themselves. This is what helps to provide a means of deflecting that famous charge of inconsistency: that Kant has no right to say anything at all about the substances that are things in themselves. When Kant says that we can have no knowledge of things in themselves, he means that we cannot make use of the pure concept of a substance in a manner that will enable us to determine a thing 'through distinctive and intrinsic predicates'. It. is compatible with this that one can use the pure concept in a manner which will allow one to assert the existence of substances, and to assert that they must have intrinsic properties: for this use falls short of a use that attempts to determine a thing by ascribing to it particular distinctive and intrinsic predicates. The schematized concept of substance, in contrast to the pure concept, is one that we can unproblematically put to use in the field of experience. What is central to the schematized concept of substance is the notion of an enduring substratum of change. In the First Analogy Kant says: In all change of appearances substance persists, and its quantity in nature is neither increased nor diminished. All appearances contain the enduring (substance) as the object itself, and the transitory as its mere determination, that is, a way in which the object exists . . . The enduring is substance in the [field of] appearance, that is, the real in appearance, and as the substratum of all change, remains always the same. Since it cannot change in its being; its quantity in nature can neither he increased nor diminished . . . In all appearances the enduring is the object itself, that is the substance as phenomenon. (Ai82/B224, B225, A183/3227)

Phenomenal substance is something in the field of appearance: the substratum of change, the bearer of changing properties, the 4reaP in the field of appearance, matter, something whose quantity is conserved. The schematism of the pure concept of substance yields not simply the abstract notion of a something that can be thought as a subject, but the notion of an enduring something that can be thought as a subject. What is this enduring something? On the assumption that the First Analogy is compatible with Kant's dynamical theory, then matter, the phenom2 This is at least one sense of 'determine' to be found in Kant. To determine a thing completely would be this, according to the 'principle of complete determination*: for each given thing, if all the possible predicates of tilings in general are listed together with their oppositess to ascribe to the thing one from each pair of contradictory opposites (Asy^/B&Go),

Substance and Phenomenal Substance

51

enal substance of the First Analogy, is force. The conservation, principle of the First Analogy thus comes close to a principle of the conservation of energy, as Guyer suggests.3 Kant says, recall: All, that we know in matter is merely relations (what we call its intrinsic properties are intrinsic only in a comparative sense), but among these rdatwm some it-re,.. enduring. (A.285/B34I, emphasis added) Kant describes the relations as forces of attraction and impenetrability. And in the Anticipations of Perception (to be discussed in Chapter 8), Kant says that 'the real* is force. What endures in the field of appearance is force, Kant, according to Guyer, 'employs endurance as the primary criterion of substance-—indeed, perhaps even endurance, of action* * Guyer sees close interconnections between the analogies, in particular between the concepts of phenomenal substance and causality, and he cites a neglected discussion of substance in the Second Analogy:5 The concept of causality leads to the concept of action, this leads to the concept of force, and thereby to the concept of substance . . . I cannot leave unconsidered the criterion of substance in experience, in so far as substance seems to make itself manifest not through the permanence of appearance, but better and more obviously through action. Where there -is action, mid therefore activity andforce^ there is also suffstance . . . How can one infer directly from action to the endurance of action? For endurance is the fundamental characteristic mark of substance (as phenomenon) , . . Now force proves action, which is, in experience, the adequate criterion of substantiality. ^204/6249-^205/6250, emphasis added)6

Concepts of causality, action, force, substance are here seen to be inextricably connected. Kant says 'where there is action, and therefore.,. force, there is also substance1. We can take Kant to be making an identity statement here: enduring action, or force, is substance (as phenomenon). This suggestion makes good sense of the assumption that the matter of the First Analogy is also the matter of Kant's dynamical theory, and we shall see that there is still more to be said in its favour. 3

4 s

Guyer, Kant and ihe Claims, 233.

Ibid. 233, I say neglected because Kemp Smith is dismissive ('not of any very real importance . .. we may omit all treatment of it', Commentary,, 380); so too is Allison (it is 'elliptical', Kant's Tramcendenud Idealism, 214), despite the fact that this passage appears to undermine Allison's own interpretation of substance as a merely spatial enduring thing; II. J, Paton expresses deep uncertainty about it (Kant's Aletaphysic of Experience (London.: Allen & Un win, 1936), 216 n. 6). A very good discussion, on the other hand, is to be found in Gordon Brittan, Kant *.« Theory of Science (Princeton, N.J.: Princeton University Press, 1.978), * The translation is guided by Guyer's suggestion, and differs markedly from that of Kemp Smith, who inexplicably omits the *Kmft ! of the last sentence.

52

Substance and Phenomenal Substance

So far we have two broad conceptions of substance; the pure concept of substance, and the schematized concept of phenomenal substance discussed in the First Analogy. Allison draws attention to these two notions of substance in a passage which aims to defend Kant from the criticisms his First Analogy arguments have attracted, Allison, says that, far from merely conflating a number of distinct conceptions of substance, Kant in fact draws on a relatively univocal conception of substance that is 'more or less common to the Western philosophical tradition5. Substance, on this conception, is regarded as a subject of predication or hearer of attributes that cannot itself be home by anything else [and] as an enduring substratum of change. These two characterizations are of course not equivalent. . . Nevertheless, they can be said to constitute two sides of a conception of substance that is more or less common to the Western, philosophical tradition, to which Kant is obviously an heir. Indeed, the subject aspect of the conception is reflected in his nominal definition of substance, and the enduring substratum aspect is reflected in his characterization of the schema?

By 'nominal definition* of substance Allison means what Kant calls the 'pure concept7 (A 147/6186). Allison thinks that these two conceptions of substance are brought together in the First Analogy, According to the Distinction however, substance is not in the field of appearance, but is the thing in itself, This means that there must be a sense in which Kant thinks phenomenal substance is not a substance: that the enduring substratum is not a first subject, and does not conform to the pure concept (or 'nominal definition*). This has implications for the debate about the First Analogy, although it is not my task here to spell out those implications. For if the argument of this chapter is correct, then far from being, as Allison says, an 'heir* to this Western philosophical tradition, Kant is a drastic departure from it. Both characterizations of substance are to be found in Kant's philosophy, as Allison rightly says; but the enduring substratum of change is not a bearer of attributes that cannot itself be borne by anything else. Matter is not a first subject.

7 Allison, Kant's Transcendental Idealism, 21:4, emphasis added. The chief accusation levelled at the First Analogy is the alleged illegitimacy of the move from a bearer of properties to something sempiternal. (See e.g. Bennett, Kant's Analytic, 197—201; Strawson, Bounds of Sense, 125—32). Gordon Brittan also thinks there are two conceptions of substance at work in the First Analogy, the Aristotelian (which is supposed to combine the ideas of first subject and substratum of change) and the Cartesian (the idea of a self-subsistent thing). As for as I can see, what Brittan describes as the Cartesian is part of the, pure concept of a first subject, I agree with Allison that these distinctions are less important than the distinction between the pure and the schematized concepts. SeeBrittan, Kant's Theory of Science, 1:43-4,

Substance and Phenomenal Substance

53

3. The Concept of Phenomenal Substance in General Implicit in Kant's label of 'phenomenal substance' is the implication of something that is not a substance. There is a general conception of phenomenal substance that has a significance in Kant's own metaphysical background, quite apart from the particular content he wants to give it in the First Analogy, In general, and quite apart, from any distinctive theses of the Critique of Pure Reason, it is possible to give a tolerably clear explanation of what it means to call something a phenomenal substance. I call it a concept of phenomenal substance in general because it is quite abstract, and because in describing it one does not need to mention time, as one does for the schematized concept of the First Analogy. It can be drawn from the passages that follow.8 Kant says that a phenomenon that is a substratum for other phenomena is not thereby a substance, or is only a substance comparatively speaking. (R 5312)9 Kant here describes a certain notion of phenomenal substance to be found in the textbook he used to teach metaphysics, written by Baumgarten: § 193. If accidents are viewed as subsisting in. themselves, they are phamomma- ml>stanliaM.w When accidents are viewed this way, they are so to speak 'substantiated' or reified—hence the Latin label of phaenomma mhsta-ntiata^ phenomena that are made substance. This is what I shall here call the notion of phenomena! substance in general, Kant has these marginal comments on Bauntgarten's definition: 'A real subject is substance. An accident can be a logical subject' 8 I first learned of this general conception of" phenomenal substance from Ermanno Bencivenga's reference to Baumgarten's definition, quoted below. See Bencivenga, Kant '$ Copermean Revolution^ 33. He emphasizes the influence of Baumgarten's Aielaphysica on Kant's thought, but thinks Kant disagrees with Baumgarten, retaining none the less much of his terminology: 'many of the obscurities of the Critique can be explained (away) as the result of forcing Kant's new wine into Baumgarten's old bottle* (224—5 n. i). Although I disagree with Bencivenga's interpretation of Kant, I have benefited from his interesting and wide-ranging discussion., He draws attention to the basic conceptual distinction, in Kant/Baumgarten metaphysics between internal and external determinations of substances, but offers a different suggestion about Its implications for Kant's mature philosophy. On the assumption that cognitions are determinations of a substance, the question arises: are they internal or external determinations? According to Bencivenga, the fact that neither answer is satisfactory shows that 'the old conceptual framework* has no adequate account of knowledge, and this is part of what provoked Kant's 'Gopernican revolution' (eh. 2, sects, i and 2).

Ak. xviii. 150, M 1777-80. 10 *§ 10,3, Accidentia si videntur per se subsistentia, sumphaemmena substantiate*, Baumgarten., Metaphysics Ak, xvii. 67, Sec also § 333, Ak, xvii, 78,

54

Substance and Phenomenal Substance 11

(R 3573? 3574)- The concept of a phenomenal substance in general. Is that of something that can be thought of as a substance, which is ultimately not a substance, We can treat things that are not substances as if they were. That is what we do when we view something that is not itself a substance as ca substratum for other phenomena'. That is what we do when we view 'accidents* as 'subsisting in themselves1. That is what we do when we regard an 'accident' as a "logical subject' of some other predicate. Kant describes this broad use of the concept of substance in the Critique: *I can say of anything and everything that it is a substance, in so far as I distinguish it from mere predicates and properties' ^349). I can treat anything at all as a substance— as a logical subject—if I treat it as though it were distinct from mere predicates and properties. This general conception of phenomenal substance is to be sharply distinguished from the pure concept of substance. The pure concept of a substance is of a 'something which can be thought only as a subject, never as a predicate of something else1 (A 147/8186); 'substance is that which is . . . an absolute subject, the last subject, which does not as predicate presuppose another further subject' (R 5295). The general concept of a phenomenal substance, by contrast, is of a something which can indeed be thought as a subject, but as a subject that must in turn be thought of as a predicate of something else. This notion of a something treated as substance, that does not really deserve the name, is associated with the labels of comparative substance, or phaenomena substantinta, Kant says that the poets reify qualities in this manner. He says that 'all properties that the poets personify are phaenomena substantiate.12 Shakespeare says that love is a spirit all compact of fire, that love is blind, that love is too young to know what conscience is. Love is spoken of as if it were a thing, indeed as if it were a person. Someone might also say of a battle, that it was fierce, and treat the battle as a subject of a predicate. One might say of a battle, that it lasted for three days, and that two brothers were killed in the same battle. One might think of the battle as if it were a comparatively self-subsistent individual thing. But battles are adjectival on the existence and actions of soldiers, and if we reify them they are merely phenomenal substances. One can say of a rainbow that it is bright and beautiful, and treat the rainbow as a subject of certain predicates. The rainbow is, so to speak, a 'phenomenon that is a substratum for other phenomena': for although one treats the rainbow as the subject of certain predicates (hence as 'substratum for other phenomena*), it is in turn adjectival on something else, the drops of rain, and so itself a phenomenon. Raindrops in their turn serve as the 11

AL xvii. 67, A 1769-71? '•"AL xvii. 291, A 1768?

Substance and Phenomenal Substance

55

'substratum' for the phenomenon' of the rainbow, which may be what Kant means in the Aesthetic when he says; 'The rainbow in a sunny shower may be called a mere appearance, and the rain the thing in itself (A^/Bh^). Raindrops themselves are bodies, and are in turn adjectival upon something else, namely matter, constituted by the fundamental forces. Something other than a body serves as the substratum for the phenomenon of body. That is why Kant says; Body . . . is a phaenotnenon subslunliatum, Bodies are the phenomena of external presences [i.e. forces]. Bodies are comparative substances, substrata phaeno-menorum [i.e. substrata of phenomena]. ({£4421, R4422,115294)13

Bodies can be viewed as substances, because they can be viewed as the bear™ ers of properties: a house is large, a ship moves downstream. They can be viewed as self-subslstent in so far as they persist for a certain, if finite, length of time. Bodies, such as raindrops, can be the substrata for other phenomena, such as the rainbow. However, bodies are not substances, since they are adjectival on something else: bodies are the phenomenon of matter, or forces. In so far as they are things that are thought of as if they were substances, without really being substances in their own right, bodies are pkaenornena substantia-ta,. In all of these examples we have something that is not a true substance, that can nevertheless in certain circumstances serve the purpose of substance. There may be different reasons for supposing that the thing in question is not truly a substance: perhaps it is because it is a property or relation (love); perhaps it is because it is reducible to, or supervenient on, other particulars and their properties or relations (a rainbow, a battle). There may be different ways in which the thing that is not a substance is treated as if it were one: perhaps the thing is treated as the logical subject of a predicate; perhaps it is viewed as a bearer of properties, or a relatively enduring bearer of changing properties. There may be different degrees of seriousness with which the thing that is not a substance is viewed as if it were one, (It is no surprise that we never see love walking with a white stick. It can be a surprise that we never reach the foot of a rainbow) The broad notion we are considering is insensitive to these differences: in all these cases, whatever the differences, we have something that is not a true substance, that is in some sense treated as if it were one, In these examples we have things—love,

1S Ak. xviL 540, M 1.773; Ak. xviii, 145, M 1:777-80, Compare Bennett's discussion, of substance as category of subject predicate statements: 'anything counts as a substance which can be referred to by a subject term*, for example 'His amiability cloys' (Kant !s. Atialylic, i 83). The amiability of Bennett's example is a pha-enomenan sub&iantiat-um, in Kant's* sense.

56 Substance ami Phenomenal Substance rainbows,, raindrops, battles—that conform to what I have called the concept of phenomenal substance in general. They are phaenomena substantiata, though none mentioned so far are phenomenal substance in the sense of the First Analogy, since none endure, We now have three conceptions of substance before us; (i) the pure concept of substance as an absolute subject, (2) the schematized concept of substance as the enduring subject of change, associated with the First Analogy, and (3) the concept of a phenomenal substance in general, a merely comparative subject. The third of these notions is defined by contrast to the first. Our question concerns the status of the second: the status of phenomenal substance as an enduring subject of change. We are now in a position to see how it might be possible for enduring phenomenal substance not to be a true substance. In the section that follows T shall try to show that substance in sense (2) is an instance of substance in sense (3): the enduring in appearance is, according to Kant, a merely comparative subject, a pha-enomemm substantiation* Substance in the First Analogy sense is something that is not substance, that can none the less be regarded as substance. The substance which is 'the real in the appearance', which persists through ail time, whose quantity is conserved, and which Kant says we call 4niatter*, is also phenomenal substance in the general sense described in the quotations above. That is to say, it is 'a phenomenon that is a substratum for other phenomena- and 'not thereby a substance, or only a substance comparatively speaking* (R 5312), It is not a substance, because it is a property of a, substance. It is an 'accident* that is viewed as subsisting in itself It is an accident that is viewed as a subject (R 3573-4). It is indeed 'a something that can be thought... as a subject'; but not 4a something that can be thought only as a subject, never as a predicate of something else* (A147/B186, emphasis added). 4, Matter as a Merely Comparative Subject First, it can be admitted that matter is indeed something that can be thought 4 as a subject'. This can be freely acknowledged, for this is part of what it means for something to be a phenomenal substance in the general sense. Matter can be treated as a subject of certain predicates: parts of it are heavy and dense, and arranged in structures that we call raindrops and houses and ships, all of which are adjectival on matter in the way that battles are adjectival on soldiers. Matter is 4an abiding appearance in space (impenetrable extension) [and] can . . . be the primary substratum of all outer perception* ^284/6339). The endurance of matter distinguishes it from other phenomena mb&iantmta^ the rainbows and raindrops, for it enables matter not only

Substance and Phenomenal Substance

57

to be thought as a subject, but to serve the task of a substratum that underlies all the changing properties with which we can become acquainted. Kant says in the First Analogy: All appearances contain the enduring (substance),,. and the transitory as its mere determination . , , The enduring is substance in the |fieldof] appearance, that is, the real in appearance, and as the substrate of all change, remains always the same, (Ai 82/6225)

When particular bodies change, when logs are heated, glow, and disappear in flames and smoke, matter itself persists through the changes. Transitory things are determinations of substance: transitory things are temporary arrangements of matter, which itself endures through all change. Matter is the subject, to which a changing series of predicates are attributed. The endurance of matter guarantees that matter will always be available to serve the role of substratum of any changing property. It gives matter a special status, among phaenomena mbsta-ntiata. Kant comments on this special status in the second Paralogism, where his purpose (which does not concern us here) is to deny that we have any notion of substance that can be used to yield knowledge of properties of the soul as a simple substance. He says So far from being able to infer these properties [i.e. of a soul] merely from the pure category of substance, we must, on the contrary, take our start from the endurance of an object given m experience, if we want to apply the concept of substance in a manner that is empirically serviceable. (A34§)

Kant says here that matter is the only thing to which we can apply the concept of substance in a manner which is 'empirically serviceable', and it is empirically serviceable precisely because it endures, Although matter can serve the function of a subject, note the apparent reservations m Kant's description of matter, even in such passages as these. He speaks here and in the First Analogy as if we must find something that will serve the function of a substance, as a bearer of properties, and the only thing available to us to serve this function is matter. In matter we find something *in the [field of] appearance which we name substance', we *give an appearance the title "substance" just for the reason that we presuppose its existence throughout all time1 (A 185/11228). Matter is something which we name substance, something to which we give the title 'substance', something to which we can apply the concept of substance in an empirically serviceable manner. All these ways of speaking suggest that matter, the phenomenal substance of the First Analogy, is something that can do the job of substance well enough; but they express caution. The reasons for caution, become evident when, we see that Kant in fact denies that matter is substance. He explicitly denies in the Critique that

58

Substance and Phenomenal Substance

matter, the enduring subject of changing predicates in the field of appearance, is a 'something which can be thought only as subject, never as a predicate of something else'. He explicitly denies that matter is 'that which is . , . an absolute subject, the last subject, which does not as predicate presuppose some further subject', Matter, he says, is substance in the appearance, and 'substance in the appearance... is not an absolute subject* (A.$2 5/13553), This admission means that the schematized concept of substance, notwithstanding its empirical serviceability, falls short of the requirements given by the pure concept of substance. The pure concept is a concept of something that is an absolute subject, which is not a predicate of something else; the schematized concept, of the First Analogy, is not. The denial that matter is substance is to be found in a variety of Kant's writings, and is a common thread throughout different stages of his thinking. Matter is no substance, but only a phenomenon of substance. That which endures in appearance, what lies at the basis of the manifold in bodies, we call substance. We find in bodies substances which we call substances only by analogy,14

This discussion assumes the dynamical theory of matter, according to which matter is constituted by forces of attraction and repulsion, that are not themselves substances but rather relational properties or powers of substances. Matter is 'that which persists in appearance, what lies at the basis of the manifold in bodies7. Matter is thus the phenomenal substance of the First Analogy, that which persists, and is the 'substratum' of the various changes that bodies undergo. Matter is that which Sve call substance'. Matter serves the task of being a subject of (changing) predicates, and to that extent it is appropriate to call it substance. The reservations implicit in this way of talking are here made explicit. Matter is not truly substance, but *only a phenomenon of substance'. Matter can be thought of as substance, but only 'by analogy'. Here Kant says that matter is something that we «i//substance, that is in fact only a phenomenon of substance: a description that conforms precisely to his description of a phaenomenon substantiatum. Compare the immediately preceding discussion with some of the passages from the Amphiboly: All that we know in matter is merely relations (what we call its intrinsic properties are intrinsic only in a comparative seme'),, but among these relations some are . , . enduring, and through these we are given a determinate object. ^285/6341, emphasis added) 14 Guyer gives a tentative date of 1778. The passage is from Mctaphysik L/, Ak, xxviii, 209, cited and discussed in Guyer, Kant and the Claims, 234—5. Guyer offers a different interpretation.

Substance and Phenomenal Substance

59

An abiding appearance in space (impenetrable extension) can. contain only relations and nothing at all that is absolutely intrinsic, and yet be the primary substratum of all. outer perception. ^.284/18339) Compare also the following passages, which I quoted only in part above. Body is not composed of substances, but is a phenomenon subslanrialum. Bodies are the phenomena of the external presences [i.e, forces] of, . . substances. (R 4421, R 4422, final emphasis added)1-"'

Kant does not always seem to distinguish very clearly between bodies, particular ephemeral, arrangements of matter, and matter itself, the enduring, But both alike count as mere phaenomena substantiates. This is true of bodies, since they have matter, or force, as their substratum. However, what is true of bodies is also true of matter itself,. Bodies are a phenomenon, of matter, matter (force, 'external presence') is in turn a property of something else, in which case matter itself is therefore substance only 'comparatively speaking'. This understanding of phenomenal substance finds clear support in many passages in the Reflexionen, of which the following are a sample. AH external appearances are phaenomena substantiata, for we treat them as substances. (R 4494) Bodies, that is to say, external appearances, are phaenomena substantiate; that is to say, enduring substrata of other phenomena. (R 4495) Bodies are phaenomena substantiate- . . , because they are external appearances, in which the first subject is absent. (R 4699} In the field of phenomena one can certainly come upon comparative substance, and phaenomena substantiala. But one cannot coroe upon the substances themselves, (R 4830) The substantial is completely unknown. (R 4054)16

Although Kant sometimes refers to matter as 'bodies* in these passages, the reference to endurance, and to the realm of appearance in its entirety^ suggest that it is strictly matter, substance of the First Analogy, that he has in mind. Here we have the same themes: something in appearance which is not a substance, but treated as a substance, something that is a substratum of other phenomena, but not itself a first subject, only a comparative subject. 15 Ak. xvii, 540, M 1772. That Kant means 'force' by 'external presen.ce' becomes clear in the discussion, of Chapter 5 (the terms are used synonymously in the Physical Monadttbgy). U} Ak. xvii. 572, M (773; Ak. xvii. 573, M 1773; Ak. xvii. 679, M 1775; Ak. xvii. 740, M I 775i A k - xv »- 399, A 1769-70.

60

Substance ami Phenomenal Substance These passages help os to understand what Kant means when he says in the Critique:. If, in thought, we were to take away all compositeness from matter, nothing at all would he left. This does not seem compatible with the concept of a substance, for a substance is really supposed to be the subject of all composition, and ought to remain, as elements, even when the connection in space through which it constitutes a body is taken away This would be true if we were thinking of a thing in itself, through pure concepts. But it does not hold for what, in the appearance, we call substance. For this is not an. absolute subject, but only a sensible abiding picture [of an absolute subject], (A.S35/B553, emphasis added)

Kant says here that phenomenal substance 'is not an absolute subject, but only a . . . picture'. How are we to understand this way of speaking? It has the form: not an /% but only a picture. If I were to say, 'this is not a real flower, but only a picture', I would say something with a similar form: and the implication would be, merely a picture of a flower. When Kant says that phenomenal substance 'is not an absolute subject, but only a . , . picture', the implication is similar: merely a picture of an absolute subject. Kemp Smith translates the phrase berharrliches Bild clef Sinnlichkeit as 'an abiding image of sensibility', but this is ambiguous between two thoughts; the first about how phenomenal substance is presented to us (via sensibility?); and the second about what phenomenal substance is an image or picture of (of sensibility?). The most plausible reading of the passage is not the latter, but the former, hence my disambiguation. Phenomenal substance is something abiding that is presented to our sensibility (and therefore of sensibility, sensible); and it is a picture, or image, of an absolute subject, hence my bracketed interpolation. This is in keeping with the usages we have noted: matter is something we 'call substance*, but it is substance only 'by analogy', 'comparatively speaking'. The pure concept of substance is the concept of an absolute subject, something independent of its relations and connections to other things, hence something that would remain even if all the things with which it was connected were taken away The pure concept of substance is the concept of a thing that is able to exist on its own, something whose existence is compatible with loneliness. Kant says here that matter does not conform to this pure concept of substance. Phenomenal substance does not conform to the pure concept of a substance, but it serves the task of a substance as far as we are concerned, because of its permanence, or abidingness. The notion of its being a 'comparative substance' is expressed with the metaphor of the 'abiding picture' (beharrliches Bild): matter is a mere picture of substance which can serve as a picture because of its abidingness. Matter is an. image of an absolute subject, and despite its shortcomings, it is the best available to us, given that we must have a subject that can be given to the senses.

Substance and Phenomenal Substance

61

Substance as such Is not to be found in matter or in the appearances. This is stated explicitly just after Kant's definition of a merely comparative substance, written close to the period of the Critique: A phenomenon that is a substratum for other phenomena is thereby not a substance, or only a substance comparatively speaking. In the appearances we cannot be acquainted with something as substance, (J? 5312)1' Knowledge of appearances does not give us any way of being acquainted with something as substance, This is precisely what is being asserted in the passage from the Amphiboly considered in the last chapter. Matter is substantial pha-enomenon. I search for that which belongs to it intrinsically in all parts of the space which it occupies, and in all the actions it performs, and these of course can only he appearances to outer sense. So 1 have nothing that is absolutely intrinsic, but only what is comparatively intrinsic, and that is itself again constituted by external relations. It is silly to suppose that matter has an absolute intrinsic nature of the sort conceived by pure understanding, for matter is not an object of pure understanding. On the other hand, the transcendental object which may be the ground of this appearance that we call matter is a mere something. ^277/0333) Kant says that 'it is silly to suppose that matter has an absolute intrinsic nature of the sort conceived by pure understanding'. Matter does not conform to the pure concept of substance, for it lacks intrinsic properties. He goes on to say that although it is silly to expect matter to have an intrinsic nature, as if it were something that conformed to the pure concept of a substance, there is a temptation to think what is given to us in experience must conform to the pure concept. The temptation should be resisted: ! must not say the following: *A thing cannot be represented through bare concepts without an absolutely intrinsic nature. Therefore it is true of both things in themselves (to which these concepts apply), and the intuition of those things, that they possess nothing extrinsic ungrounded in something absolutely intrinsic', (A284/B340, italics added)18 We must not infer that the pure concept of substance, which applies to things in themselves, also applies to the intuition of those things, i.e. to phenomenal substance. Hence we must not infer that; since the pure concept of substance requires the presence of intrinsic properties, things in themselves and phenomenal substance both have intrinsic properties. The pure concept does 17

Ak, xviii. 150, M 1:777 80.

is fviy translation differs significantly from Kemp Smith's; *so kann ich nicht sagen: wdl, ohne ein Schlecthininneres, kein Ding durch bloBe Begriffe vorgesfellt wcrden kann, so sd auch in den Dingen sdhst, die unter diesen Begriffen enthalten sind, und ihrer Anscbauung nichts AuBeres, dern nicht. etwas SchlechthininnerHches ssum Grunde lage1 ^284/8340).

62

Substance and Phenomenal Substance

apply to things in themselves, but not to the intuition of those things, i.e. phenomenal substance; things in themselves have an absolute intrinsic nature, but phenomenal substance does not. There are striking similarities between these passages from the Critique, and the passages from outside the Critique just considered, Kant says repeatedly that matter is substance only 'comparatively speaking'; in the Amphiboly he says that we are acquainted with properties that are only 'comparatively intrinsic'. In the earlier cited passages Kant speaks of 'substance' of which matter is 4the phenomenon'; in the Critique he speaks of the transcendental object which is 'the ground' of 'this appearance that we call matter'. That ground is 'something of which we should not understand what it is, even if someone were in. a position to tell us?; it is the 'substratum [of matter] which cannot be known* (.^,359), In short, the contrast between matter and substance in Kant's other writing is present in the Critique as the contrast between matter and things in themselves. The reason that we cannot be acquainted in the appearances with something as substance is that we can be acquainted only with forces. Although forces can constitute matter, forces are not substances. The passages which follow make it evident that Kant thinks we are acquainted with force, not substance, and they shorn- why he thinks that force and substance are not the same. The first, from the Amphiboly, has already been encountered. The others show how it is to be understood. We are acquainted with substance in space only through forces which are active in this and that space, either drawing other objects (attraction) or preventing their penetration (repulsion and impenetrability). We are not acquainted with any other properties constituting the concept" of the substance which appears in space and which we call matter, ^265/6321) We are acquainted with the substantial only through that which endures , . . All that we are acquainted with of substance is force, fundamental force, (R 4824)I9 In the end, however, we find everything in the object to be accidents. The first subject is something that makes the accidents possible . . . It is something with which I am acquainted through the enduring accident of impenetrability and also , . . attraction. We judge the reality of the accidents through the sensation of an object. The reality of the accidents is distinct from the reality of the [first] subject, (*44i2)20

w 20

Ak, xvii. 739, A 1776-7.

Ak, xvii, 537, M 1:773, Kant sometimes denies that force is an accident, and says instead that it is a relation, but most generally it seems to be viewed as a relational accident, or property. Compare the idea of substance as something that 'first makes the extrinsic properties possible* in the Amphiboly, AaSj/t^Q.

Substance and Phenomenal Substance

63

All three of the above make the point that we can be acquainted with substance only through forces. The first appears to treat the following concepts as coextensive: matter, substance appearing in space, properties of repulsion and impenetrability. This identification of what Kant describes as properties (forces) with phenomenal substance makes sense on, the present interpretation of phenomenal substance; something that can be treated as a substance, that is ultimately a property of something else. The second and third of the above identify that which endures as force: we are acquainted with substance through that which endures, and that which endures is force. This supports Guyer's suggestion about the connections between force, action, and substance raised in A2O476249., where Kant speaks of force, or action, as the adequate criterion in experience of substantiality.21 The third passage refers to the 'enduring accident of impenetrability'.22 It says that force is an accident, or property, of a substance. We have seen already that Kant says that matter 'consists entirely of relations': that matter is constituted by the forces of attraction and impenetrability. Here the same idea that matter is force is taken to imply that 'we find everything in the object to be accidents'. The conclusions are not so very different if Kant thinks that there is a sense in which forces are both relations and accidents: that they are relational properties, In conclusion, substance in the First Analogy sense is '&phaenommon substantiatum: a property, that is treated as if it were a substance. In the realm of appearance we are acquainted only with forces, and forces are not substances but properties, relational properties, of substances unknown in themselves. Since matter, or the forces that constitute matter, is the substratum of other phenomena, matter is a non-substance that occupies the role of a substance for us. It is a 'picture' of an absolute substance, that, in virtue of its endurance, fills that role very well. We can see, then, that Kant's commitment to phenomenal substance does not undermine the Distinction he holds. We can see how he can consistently believe that phenomenal substance is in the realm of appearance, while substance is not in the realm of appearance. Moreover, we can see how Kant might believe that despite our ignorance of substance, we are required to believe in its existence. This brings us back to a topic raised at the beginning of this chapter. 21 It also supports Guyer's suggestion that Kant's views on the endurance of force imply a principle about the conservation of energy rather than of matter as traditionally conceived. Guyer, Kant and the Claims^ 233. 22 This is implicit in the Critique in other passages too; the enduring is, according to the First Analogy, 'the real in appearance' (As 82/1^225), and the real in appearance is, according to the Anticipations of Perception, force (A 168/6210 IT.), which again favours Guyer's suggestion that substance is 'enduring action* or force.

64

Substance and Phenomenal Substance 5. Note on the Inference to Substance

Kant says that the first subject, or substance, is what makes the accidents possible (R 4412), If there were no substance, there would be no accidents, This requirement that there be a bearer, or subject, for accidents, resembles the requirement raised at the beginning of Chapter 2 that there be a bearer, or subject, for relations: that 'concepts of relation presuppose things which are absolutely [independently] given, and without these are impossible* (A284/B34O). Indeed if the 'accidents7 in question are relational properties, it may be that the requirement of a bearer of ^accidents' amounts to the same thing as the requirement of a bearer of relations. Just as Kant said that the bearer of relations must be thought of as having some independent existence, some 'intrinsic nature which is free from all external relations' (A274/B330), so he says, in similar vein, that the 'reality' of the first subject is distinct from the 'reality' of the accidents of impenetrability and attraction. It is not enough that we are able to treat these latter accidents or relations as i/they were substances. We are always able to treat any non-substance as if it were a substance, a ph&enomenon mlntantiatutn^ if we need to, I can say of anything and everything that it is a substance,, in so far as I distinguish it from mere predicates and properties. (A349) Everything can be thought as either a subject or a predicate, but not everything can exist as a subject. (R $$$$)23>

It is not enough to be able to think of something as if it were a subject: we are required to think that something exists as a subject. Since logical, subjects can always be predicates, we are required by laws of reason to think a last subject. This is substance. (R 4052) The concept: of substance is a terminating concept, (R 403 g)24

Matter can be thought as a subject, but it cannot exist as a subject: not because it is dependent on our minds, but because it is adjectival in its character. Confronted in experience with predicates, or properties, or relations, we are required to think that they have bearers that have some independent existence. Confronted in experience with forces, even enduring forces, we are required, given (in Kant's opinion) the relational character of forces, to think they belong to something else. If one knowrs that the thing one takes Z3

Ak. xviii. 370, M 1780-1,, -4 Ak, xvii. 398-1), A 1769-70; Ak. xvii, 394, A 1769-70, Guycr comments on. the use of

substance as a terminating concept, Kant a-nd the Claims, 234—5.

Substance and Phenomenal Substance

65

for a substance Is mere pha-enomenon substantiatum^ one must infer the existence of something else. Hence there must be substances, things that conform to the pure category of substance, that are absolute subjects, capable of independent existence, As substances that conform to the pure category; they have an intrinsic nature that is free from all external relations; they have intrinsic properties ^274/8330). As bearers of relational accidents they have a reality that is distinct from those accidents (R 4412). Their relational properties, forces, endure, affect us, and appear to us in space. There is further support here for the explanation given in the previous chapter of why Kant thinks it so obvious that the notion of appearance logically implies a thing in itself: that appearance implies a something that appears. If matter is 'constituted entirely of relations', if it is nothing but relational 'accidents*, then matter in. turn must, have a bearer, a substance that, unlike matter, is an 'absolute subject'. And this is why Kant says that 'matter is mere external appearance, the substratum of which cannot be known through any predicate that we can assign to it* (A359). Kant would not see this as an illegitimate use of the pure concept, since, while asserting the existence of a substratum, it also denies that we thereby determine a thing 'through distinctive and intrinsic predicates5 ^565/6593). Without having access to the properties that 'belong to the thing whose role is to be a first subject' (A 147/13186), we cannot use the pure concept in a manner that enables us to determine an object. A similar thought may also be present in the following (admittedly cryptic) passage. Substantiality and its opposite: mere relation. [Substantiality] is not to be seen . .. because the predicates are lacking. (R 4493)25

We cannot 4see' substantiality because the appropriate predicates, intrinsic predicates, are lacking. The alternative meaning of the notion of 'external appearance* recurs in the passages we consider here. Kant says (A359) that matter is an 'external appearance* of a substratum. By external appearance Kant does not mean that matter is a construction of sensations in the spatial array provided by outer sense. If he did, then Bennett's accusations of illegitimate conflation of sensations with properties of substances would be quite justified. 'External', here, does not in the first instance concern space, nor, here, does Kant even mean to emphasize that matter is outside as, although it is. 'External*, both here and in these other passages, is to be understood relative not

25

Ak. xvii. 572, M 1773. Substance is one of three 'terminating concepts' in the discussion, I have extracted, the part about substance because it seems to have much in common with the other passages cited.

66

Substance and Phenomenal Substance

primarily to us but to the substance or substratum. This usage of 'external appearance1 occurs in the quatrain of Reflexionen considered in Section 4: All external appearances are phaenomena substantiality for we treat them as substances. (R4494) Bodies, that is to say, external appearances, are phuemtmemi mbsl-anliala^ that is to say, enduring substrata of other phenomena. (R 4495) Bodies are pkaenomena mbstantiata . . . because they are external appearances, in which the first subject is absent. (R 4699) In [the field of] appearance one can certainly come upon comparative substance, md phenomena suhsiantiala. But one cannot come upon the substances themselves, (J?483o) Bodies are external (extrinsic) to a substance, a first subject. Matter is force, and force is an 'external' property, an extrinsic or relational property, of a substance, whose intrinsic properties, on Kant's view, are unknown. To say that matter is external appearance is to add that it is constituted by forces that are presented to m, and with which we can become acquainted. 6. Concluding Remarks and Nem Business Kant believes that there are things in themselves, that is to say, substances, They have intrinsic properties, and we do not know what those properties are. I have attempted, in this chapter, to show how it is possible to ascribe to Kant these views and at the same time recognize his commitment to the phenomenal substance of the First Analogy. In the process I have offered more support for the Distinction, and for the thesis of Humility, since most of the passages considered identify substance with the unknowable thing in itself, ascribe to substance unknowable intrinsic properties, contrast substance with its relations and relational accidents, and identify those relational properties as forces. Moreover, these passages show that Kant uses 'phenomena' to refer, not to sense data, as a phenomenalist might use the term; nor to observable events or regularities, as a philosopher of science might use it; but rather to things that are not substances. That, in the quoted passages, is primarily what the notion means, Kant contrasts substance with phenomenon. He contrasts substance with 'its opposite: mere relation' (J? 4493). He says, in the Critique, that 4the understanding calls an object in a relation mere phenomenon* (13307). The concept of a phenomenon is not a concept that, in the first instance, has to do with how things look. Nor does it, in the first instance, carry any implication of ideality. On the contrary, Kant describes relational accidents as having a reality that is distinct from the reality of the

Substance and Phenomenal Substance 67 subject (R 4412). What this means exactly will be a question for later: but for the moment the point to emphasize is that 'relational accidents' are both phenomenal, and real. Allison said that there is conception of substance that is *more or less common to the Western philosophical tradition'. It has two sides; substance as a bearer of attributes that cannot itself be borne by anything else, and substance as an enduring substratum of change. Allison said that Kant is obviously an heir to this tradition. If the argument of this chapter is correct, then Kant is not an heir but an exception to this tradition. To be sure, phenomenal substance is a substratum of change, and a bearer of attributes. But that can be true of mere pkaenomena substantiate love, and battles, rainbows,, and raindrops. To be sure, phenomenal substance is, unlike these, an all-enduring substratum of change, That indeed makes it empirically serviceable, and enables it to fulfil the task of substance in appearance. It is no first subject, though, but a property borne by something else, the 'substratum of matter\ Enough has been said, I hope, to remove the potential stumbling block posed by phenomenal substance. It is time now to turn to the question raised at the end of the last chapter, the problem raised by Strawson. Why is it that Kant believes that our ignorance of things in themselves, our ignorance of their intrinsic properties, follows from the fact that we are receptive creatures? So far we have seen many expressions of Kant's Distinction, and many expressions of Humility, but Kant's reasons for the latter are still mysterious. We have also seen that some of the clearest expressions of Humility are to be found in the Amphiboly of the Concepts of Reflection, There is something about this fact which deserves serious attention. In writing the Amphiboly Kant's chief aim is to show why Leibniz is wrong, Now, it is unlikely to be a coincidence that some of Kant's clearest expressions of Humility occur in the context of his discussion of Leibnizian philosophy. Perhaps it is even something of a clue. Perhaps there is a link between Kant's attitude to Leibniz and his belief in our ignorance of things as they are in themselves. Perhaps, then, we should consider Kant's reasons for thinking that Leibniz is wrong, if we would like to discover his reasons for Humility.

4 Leibniz and Kant I, Introduction Kant's expression of Humility is linked with a critique of Leibniz, and this critique assumes a particular interpretation of the Leibnizian philosophy;1 so if we are to understand Kant's critique we must first see what, in Kant's opinion, Leibniz is doing. The primary task of this chapter is to sketch an interpretation of Leibniz which is in accord with Kant's own interpretation, in order then to raise the question of its connections with Kantian Humility.2 Points of similarity between Kant and Leibniz will emerge, and also points of difference. The most obvious differences concern knowledgehow much we have, and how we get it. The Leibnizian philosophy is extraordinarily ambitious, without an ounce of epistemic humility. Its ambition is unhampered by a denial of causal interaction, and a corresponding absence of epistemic receptivity. However, the focus in this chapter will, be, not on differences, but on similarities., real or alleged, between these two philosophers. One point of similarity is a distinction in Leibniz between appearances and things in themselves which has much in common with the Distinction 1 I use two translations of Leibniz's works; Leihniz: Philosophic a-i Papers and Letters, trans, L. Loemker, 2nd edn, (Dordrecht: Reidel, 1969), and Lfikniz: Philosophical E$my$y trans, R, Ariew and D. Garber (Indianapolis, Ind.: Hackett, 1989); and occasionally borrow translations from Benson Mates, The Philosophy of Leibniz: Metaphysics and Language (New York: Oxford University Press, 1.986), who provides a useful 'gallery* of texts on the topic of relations (222—6). The original-language sources of works cited are chiefly Leibniz: Die Philosophiscken Schriflen^ ed. C, I. Gerhardt (abbreviated 'Gerhardt'); Leibniz: Matkemalische Schriften, ed. Gerhardt (abbreviated 'Gerhardt M'); and Opuscules et fragments inedits de Leibniz, ed, Louis Couturat (abbreviated 'Couturat'). See Bibliography for full details. 2 The interpretation offered is independently defensible, I believe, though it would be beyond the scope of this chapter to defend it in detail here, It is in broad agreement with, and has benefited from the work of, a number of different, commentators, especially Nicholas Reseller, The Philosophy of Leihniz (Englewood Cliffs, N.J.: Prentice-Hall, 1967); Mates, Philosophy of Leihniz; Fabriao Mondadori, 'Solipsistic Perception in a World of Monads', in Michael Hooker, ed,, Leibniz: Critical atuf Interpretive Essays (Manchester: Manchester University Press, 1982), 21—44; Catherine Wilson., Leibniz's Aietapkysics (Princeton, N.J.: Princeton University Press, it)8(>); and G, H. R. Parkinson, Logic and Reality in Leibniz's Mela-physics (Oxford; Oxford University Press, 1965), Particular debts and disagreements with these and other critics will be noted in due course, especially on the topic of" the rethieibilitv of relations.

Leibniz and Kant 69 I attribute to Kant. Indeed Kant's own view, with the sharp division between matter and substance described in the previous chapter, makes better sense when seen against its Leibnizian backdrop. A further point of similarity has been suggested by some commentators: that Kant endorses a Leibnizian view about the reducibility, or perhaps unreality, of relations. This further point of similarity has been taken to have implications for knowledge; a suggestion which, if true, would be of considerable interest. For while Leibniz himself is less than humble in epistemological ambition, if these critics are correct, then Kant is led to Humility, or something like it, because of his loyalty to a Leibnizian view about relations, Despite the interest of this suggestion, it is, I shall argue, mistaken. These commentators think that Humility has something to do with Kant's opinion about the reducibility of relations, and that is indeed so-—but for a reason, opposed to the one they offer. Kant advocates Humility not because he thinks that Leibniz is right about relations, but because he thinks Leibniz is wrong. That is why his expression of Humility is linked with a critique, and not an endorsement, of Leibnizian philosophy. This conclusion will take some argument, and I hope to establish it over the course of the immediately ensuing chapters. The earlier caveats about the notion of a phenomenon apply as much to a consideration of Leibniz as they do to Kant, and for the same sorts of reasons. Leibniz endorses a philosophical distinction that divides phenomena from things in themselves, and this is something he shares with many philosophers, ranging from Kant back to Plato. Here again we have the same choice between a primarily epistemological and a primarily metaphysical way of understanding a distinction of this kind. We can think that 'phenomenon' may be a label for things as they look to us, through the murky lens of sense perception. We can think that the label suggests the manifest image of the world, something to be contrasted with things as they really are, independently of our sense perception. This epistemological distinction is not quite what these metaphysicians had in mind, however, any more than it is what Kant had in mind, It is true that the realm of things that we can sense seems to coincide with the realm of phenomena for Leibniz, as for Plato before him. And it is true that we are likely to find the two realms referred to in terms of how we know them: the sensible and the intelligible. But the fundamental distinction between phenomena and things in themselves is a distinction between things that are ontologically deficient, and things that are not: it divides the insubstantial from the substantial. Thus Plato divides the realm of Forms from the realm of concrete particulars. He has a view about how we get to know each. Forms can be grasped through the intellect; particulars can be grasped through the senses. But these latter epistemological facts are not what grounds his distinction. Phenomena are deficient; but they are not deficient because they are delivered to us through

70 Leibniz and Kant the senses. Rather, the senses are deficient because they deliver us only phenomena. Leibniz's distinction is likewise, I think, fundamentally metaphysical,3 and lie goes so far as to compare his own opinion with that of Plato, Bodies arc phenomena, he says. *[B]ody itself cannot be conceived independently of other things,' and thus 'bodies do not deserve the name substance^ this seems to have been Plato's view when he remarked that they are transitory beings that never subsist longer than a moment.54 His distinction is not primarily between things as we sense them, and things as they are, but between substances and things that are not substances—regardless of how we happen to sense them. In everyday life we might think of physical things, like parrots and people, as substantial, real, basic; but things like flocks and football teams as less so. If in everyday life we think that way, we are thinking of parrots and people as things in themselves, flocks and football teams as phenomena, or at best as phaenomena substantiate. Leibniz uses a distinction between the substantial and the non-substantial to map a different terrain. Souls—monads, simple substances—are things in themselves, and bodies are phenomena. Souls are substances, bodies are not substances. There is something that the Leibnizian view has in common with everyday thinking, though. Bodies are to souls, for Leibniz, as football teams are to people, in everyday thinking. An. aggregate of substances is what I call a mfatantiatum, like an army of men, or a flock of birds, and such are ail bodies,5

2, Kant's Version of Leibniz If Leibniz believes that 'bodies do not deserve the name substances', then his belief is guided by a particular conception of what a substance is. The 3

It is also possible to have a more traditionally 'phenomenalistic' interpretation of Leibniz, as suggested, for example, by Robert Adams, who says that phenomena are primarily intentional objects of perceptual awareness (Adams, 'Phenomenalism and Corporeal Substance in Leibniz', in Peter French, Theodore Uehling, and Howard. Wettstein, eds,, Midwest Studies in Philosophy, viii (Minneapolis: University of .Minnesota Press, 1983), 217-57). This is partly a difference of emphasis. Monads have a dual function in Leibniz's philosophy, as foundational constituents of the world, and as percipients, which in turn yields two ways of understanding phenomena, as the 'results' of these foundations, or the perceptions of these percipients. For the two roles see Catherine Wilson, Leibniz's Metaphysics, 196, and Montgomery Furth, 'Monadology', in Harry Frankfurt, ed,, Leibniz (New York: Doubleday, 1972), 99-136. 4 'Conversation of Philarete and Ariste* (1712, revised 1715), Gerhardt vi. 583, 586, Anew and Garber 262 (italics added}.. It is not ephenieraltty so much as dependence that disqualifies a thing from being a substance. 5 From an undated essay, Couturat 13, Ariew and Garber 200 n. 249,

Leibniz and Kant

ji

argument for monadology takes its starting-point from what Kant calls the pure concept of a substance, and from the corresponding distinction between two classes of properties a substance may have,6 Leibniz's monadology has its sole basis in the distinction between the intrinsic and the extrinsic, which he presented purely in, relation to the understanding. Substances in general must have some intrinsic nature* which is therefore free from all external relations, and therefore also from composition .. . But what is intrinsic to the state of a substance cannot consist in place, shape, contact, or motion, since these properties are all external relations, We can therefore attribute to substances no intrinsic state except for that which we determine inwardly in our own sense, namely, the state of representations. Thus are the monads completed. They are supposed to serve as the raw material for the whole universe, despite having no active force, except for that consisting in representations (which, strictly speaking, are active only within the monads). That is why [Leibniz's] principle of the possible reciprocal, community of substances had to be a pre-established harmony, and not a physical influence. For when everything is merely intrinsic... the state o f . . . one substance cannot stand in any active connection whatsoever with the state of another. (A274/6330) Leibniz first assumed things (monads), with an inner power of representation, in order afterwards to found on these their external relations and the community of their states... The mere relation of substances was then the ground through which space was possible as a consequence; and the mere connection of [a substance's] own properties with each other was then the ground through which time was possible as a consequence. (.4267/6323). Leibniz took the appearances for things in themselves. (.4264/6320) He assumed that we intuit things as they are, although in confused representation, (Aa68/B323) The appearance was, for him, the presentation of the thing in itself. (A.27O/B32&) [Leibniz held] that our entire sensibility is nothing but a confused representation of things, that contains solely what belongs to the things in themselves^ but with a crowding together of characteristics and partial representations, so that we cannot consciously distinguish the things from one another. (A43/B6o) These passages are from Kant's critique of Leibnk in the Amphiboly of the Concepts of Reflection, except for the last, which is from the Aesthetic, I want to consider their implications for the way Leibniz at once distinguished phenomena from things in themselves, yet at the same time 'took' the phenomena for things in themselves, according to Kant. 6

This argument for monadology attributed to Leibniz by Kant iss the subject of an excellent (though I think partly mistaken) analysis by van Cleve, 'Inner States and Outer Relations',

j2

Leibniz and Kant Guided by the pure concept of substance, and by the distinction between intrinsic and extrinsic properties, Leibniz assumes that substances must have some intrinsic properties independent of any relations they bear to other things. Physical properties of place, shape, contact, or motion are not intrinsic properties, but 'external relations' of substances. However, substances must have some intrinsic properties: things capable of existing in the absence of relations to other things must have properties to match. According to Leibniz, the only possible candidates for such properties are inner representations: and thus the substances of his ontology are monads (A274/B33O), On this description, Leibniz's division between phenomena and things in themselves involves a division between relational and intrinsic properties of substances. (We may leave aside, for the moment, Leibniz's conclusion about what exactly those intrinsic properties are supposed to be.) Phenomenal properties are physical properties, and are relational properties. Properties of things in themselves are to be contrasted with those of phenomena, they are not physical, and they are intrinsic properties of substances. Leibniz seems to endorse a distinction that—at least in these last respects—resembles Kant's. A further aspect to Kant's understanding of Leibniz also emerges in these passages. Leibniz takes the appearances for things as they are in themselves (A24O/.B32O). Note that on a phenomenalistic conception of Kant's distinction between appearances and things in themselves, this suggestion is puzzling. Bennett goes so for as to say that it is, at first sight, 'intolerably puzzling',7 It seems to ascribe to Leibniz the ludicrous view that phenomena are epistemically inaccessible, as things in themselves are for the Kant of the Critique. But it is evident, I think, what Kant has in mind. There are two very clear ways in which Leibniz does indeed take the appearances for things in themselves, and they go hand in hand. Leibniz takes the appearances for things in themselves, because he reduces phenomena to things in themselves. Substances with their intrinsic properties are 'supposed to serve as the raw material for the whole universe", says Kant. On the Leibnizian account, there is a sense in which 'everything is merely intrinsic' ^274/6330), and hence a sense in which appearances are nothing over and above things in themselves. The monads and their intrinsic properties are 'grounds', or 'foundations', or 'raw material' for external relations, and the external relations of monads are nothing over and above the intrinsic properties of monads. The physical world is the world of appearance, constituted by the external relations of substances, yet relations are nothing over and above the intrinsic properties of monads. So appearances are nothing over and above things in themselves. As birds are constituents 7

Bennett, Kant's Dialectic\ 53,

Leibniz and Kant 73 or raw material for a flock, soldiers raw material for an army, so monads are raw material for the whole universe, Kant picks up this kind of metaphor at a later date, saying that according to the Leibnizian view, There is . . . no other difference between a thing as phenomenon and the representation of the noumenon which underlies it than between a crowd of nieti which I see at a great distance and the same men when I am so close that I can count the individuals. It is only, [the Leibnizian] says, that we could never come so close to it, (On a Discovery, ijgo)8

Just as a crowd is nothing over and above the men, so the phenomenal realm is nothing over and above the realm of monads. Kant says that monads are 'supposed to serve as the raw material for the whole universe, despite having no active ford. In place of an active force that would enable one substance to act upon another substance, there is a pre-established harmony of intrinsic states. The external relations of monads—the quasi-causal relations of reciprocal community operating between substances—are 'grounded' upon the intrinsic properties of the substances. These external relations in turn form the 'ground' whose "consequence' is space ^267/6323). The dynamical and spatial features of the world are thus founded upon the intrinsic properties of monads. Spatial relations are reduced to dynamical relations, and the latter in turn are reduced to harmonies, or patterns of similarities and differences, among the intrinsic properties of substances. Given the nature of Leibniz's distinction between phenomena and things in themselves, in 'taking* phenomena for things in themselves, Leibniz takes the physical realm to be nothing over and above the monadic, and he takes relational properties to be nothing over and above intrinsic properties, He at once reduces phenomena, reduces bodies, and reduces relations. Kant's observation that Leibniz takes the appearances for things in themselves has an epistemological dimension as well. In addition to their foundational role as *raw material' of the universe, monads are also percipients who perceive the world of which they are themselves constituents. Leibniz takes the appearances for things in themselves, because he believes we have perceptual knowledge of things as they are in themselves. If Leibniz takes the appearances for things in themselves, then he believes that things in themselves are perceived through the senses, According to Kant, Leibniz believes that w& can know 'through sensibility*, albeit confusedly, the nature of things in themselves (A.44/B62), Leibniz assumes that we intuit things K Ak. viii, 208, Allison 1.34-5, The *Leibnizian' in question, is Eberhard, and Kant is careful here to distinguish this interpretation of Leibniz from a somewhat different, and more charitable, interpretation, as we shall see in Chapter 9.

74 Leibniz and Kant as they are, although confusedly ^268/6323). The appearance is, for him, the presentation of the thing in itself (A270/B326). Leibniz thinks that the representations in our sensibility 'contain solely what belongs to the things in themselves' (A.43/B6o). Here we have a clear statement about what, in Kant's opinion, Leibniz believes is given to sensibility: nothing but properties that belong to things in themselves-—nothing but the intrinsic properties of monads, albeit perceived in a crowded and distorted way. Three major strands of Leibniz's philosophy have emerged in these brief reflections on Kant's interpretation of it: first, a distinction in Leibniz between phenomena and things in themselves, resting upon a distinction between intrinsic and relational properties of substances; second, a reduction of phenomena to things in themselves, resting upon a reduction of relational properties to intrinsic; third, a claim to sensory knowledge of things as they are in themselves. Let us consider these three one at a time. 3, A Distinction between Phenomena and Tilings in Themselves Things in themselves are substances, and it is part of the pure concept of substance that a substance can exist independently of its relations to other things. Leibniz interprets that independence in a distinctive way. 'Each substance is like a world apart, independent of ail other things, except for God.'9 One aspect of the independence is causal, The monads are windowless: each is a microcosm, in complete causal isolation from the others, ^Speaking with metaphysical rigour, there is no real influence of one created substance on another.110 That is why Kant says that according to Leibniz there is no relation of physical influence or 'active connection' between monads (A274/B33O). A further aspect of the independence is metaphysical: monads, unlike bodies, do not depend on other things for their existence. Leibniz says that 'bodies do not deserve the name substances', and that a true substance is not a physical thing. If (per impossibile) a substance were a body, it would have spatial parts. But nothing that has parts can be a true substance: for something that has parts depends on its parts for its existence, and is thus not an independent being. That is why 'body itself cannot be conceived independently of other things',1 T [Gjiven that a body is a whole, it depends upon other bodies of which it is composed and which make up its parts. Only monads, that is, simple or indivisible substances, are truly independent of any other concrete created thing.32

9 'Discourse on Metaphysics' (1686), Gerhardt iv. 439, Anew and Garber 47. to 'Conversation' (emphasis added), Ariew and Garber 259 n, 315 (citing an earlier draft of the work, in A. Jfobinet, Malebraneke el' Leiimtz: Relations personnelles (Paris: J, Vri.o, 1955}). 20 Letter to de Voider, 1704 or 1705, Gerhardt ii. 276, Ariew and Garber 182. -' I am not doing justice here to Leibniz's attempts to find, a coherent doctrine of corporeal substance which attributes to bodies an organic unity, so that they thereby become more than mere aggregates, though I do not see this doctrine as ultimately consistent with the monadology.

Leibniz and Kant 77 dynamical relations to each other, parts that cohere with, each other, or repel each other. The parts stand in spatial relations to other parts. Those parts will in turn consist of parts similarly related, as one descends the infinite labyrinth of the continuum. But a body is constituted by, not just physical parts bearing certain relations to each other, but at the roost fundamental level, monadic substances bearing certain relations to each, other. Bodies are dependent in two ways; they are dependent on their (bodily) parts; and they are dependent on their monadic foundations. Bodies are founded on monads, not in the way that wholes are founded on their parts, but in the way that things of one ontological domain may be founded on things of quite a different domain. We shall have more to say about this shortly. There is evidently common ground between the Kantian and Leibnizian conceptions of matter, and accordingly with their conceptions of the distinction between phenomena and things in themselves. First, matter is constituted by forces. This Leibnizian conception of matter is implicit in Kant's discussion of action and force as the criterion of material substance in the Second Analogy ^204/6249), discussed in the preceding chapter. Second., forces are construed, whether rightly or wrongly, as relations, or relational properties. Third, matter endures. Leibniz says that 'Corporeal substance can neither arise nor perish except through creation or annihilation. For when corporeal substance once endures, it will always endure, since there is no reason for any difference.122 Fourth, and most importantly, matter is something that we call a substance, and regard as a substance, that is really a phenomenon of substances. As Leibniz says, he puts corporeal forces where he puts bodies, 'namely, among the phenomena'. Corporeal substance is uphaenomenon mfatantialum: something we treat as a substance that does not, as Leibniz says, 'deserve the name'.23 Corporeal substance has a kind of inferior citizenship, a 'droit de bourgeoisie'.24 Things in themselves are substances possessing intrinsic properties. Matter is something that does not really ^deserve the name' of substance (Leibniz), something that is a mere •picture' of a substance, a merely 'comparative' subject (Kant). The properties of the physical, phenomenal world are relational. However, Leibniz says more than this. Having distinguished phenomena from things in themselves in a way that has much in common with Kant, he then 'takes1 the phenomena for things in themselves, saying that the former are nothing over and above the latter. This brings us to the question of how

22

'Primary Truths' (1686), Couturat 523, Anew and. Garber 34, italics deleted. -3 Leibniz refers to aggregates as suhstanti&ta in a number of places; see also, for example, his notes for a letter to des Bosses, 5 February 1713, Gerhardt ii, 439, Anew and Garber 200, 24 Matter is thus described in a letter to Arnaultl, 30 Apr. 1687, Gerhard! ii. 102, Anew and Garber 90.

78 Leibniz and Kant the phenomena are founded upon things in themselves, according to Leibniz. 4, A Reduction of Phenomena to Things in Themselves Leibniz believes not only that monads have intrinsic properties, but that in some sense that have only intrinsic properties: that, as Kant puts it, 'everything is merely intrinsic' (A274/B330). This is not incompatible with his belief that the phenomenal realm is constituted by the relations of substances, if Leibniz at the same time takes the phenomena for things in themselves: takes relations and relational properties to be reducible to intrinsic properties, and takes the physical realm to be reducible to the monadic. How are we to understand this reduction? Kant says that according to Leibniz, since everything is merely intrinsic, the monads form the raw material, or foundation (Grundstqff), for everything else that exists (A274/B33O), This suggests one aspect of the reduction. For any beings other than monads, there must be monads to serve as the foundation for those beings. Kant also describes the foundations as 'grounds' (Griinde) for which the other beings are 'consequents' (Falgeri) ^267/6323), This suggests a further aspect of the reduction: the intrinsic properties of monads are such that they entail the facts about the other beings. In contemporary philosophical parlance, Kant's idea is that any beings other than monads must supervene on monads and their intrinsic properties, Derivative things supervene on foundational things when they satisfy the two features suggested in Kant's description: whenever there are derivative things that are a certain way, the foundational things must be some way or another; and how the foundational things are must entail how the derivative things are. This means that there will be no difference in the derivative things unless there is a difference in the foundational things. In an article which has become the locus classicus for the topic of supervenience, Jaegwon Kirn defines the notion ntore precisely (restricting our attention to classes of properties) as follows: A class of properties A supervenes on a class of properties B if and only if, necessarily, for every object x and for every property F in A, if x has F, then there exists a property G in B such that x has G; and necessarily, for every object y, if y has G then v has F.25

35 Jaegwon Kim, 'Concepts of Supervenience', Philosophy and Phenomenolagicat Research 45 (1984), 153—76, This is the definition for what he calls strong supervenience (165), which is the only sort of supervenience we shall he concerned with here and in the chapters to come.

Leibniz and Kant

79

Apply this general Idea to the cases at hand. Suppose that the physical world supervenes on the monadic. Then if the physical world is a certain way, the monadic world must be some way or another; and the way the monadic world is must entail the way the physical world is. Suppose that relations supervene on the intrinsic properties of monads. Then, for any relation, there must, be some intrinsic properties amongst the monads; and those intrinsic properties must entail the relation. Let us turn then to Leibniz, to assess the accuracy of Kant's version of him. The idea that bodies supervene on monads, and that relations supervene on intrinsic properties, can be seen as a consequence of—or as one aspect of—Leibniz's dizzying mirror thesis, according to which everything in the universe mirrors, or expresses, everything else. In the first instance, there is a mirroring of all monads in. each monad, ordained by preestablished harmony: [TJhis mutual connection or accommodation of all created things to each other and of each to all the rest causes each simple substance to have relations which express all the others and. consequently to be a perpetual living mirror of the universe . . . This universal harmony . . . results in every substance expressing exactly all the others., .26

This mirroring yields relations of similarity and difference amongst the monads which in turn form the foundation for the realm of physical things: the physical world mirrors the world of monads, [Cjompound beings are in symbolic agreement with the simple. The souls follow their laws , . , and the bodies follow theirs, which consist in the laws of motion; nevertheless these two beings of entirely different kind meet together and correspond to each other like two clocks perfectly regulated to the same time,27

This correspondence between the two realms means that the physical realm expresses the realm of monads. Leibniz's notion of expression is broad: One thing expresses another, in my usage, when there is a constant and regular relation between what can be said about one and about the other.28 One thing is said to express another if it has properties that correspond to the properties of the thing expressed... There are various kinds of expression; for example, the model of a machine expresses the machine itself, the protective delineation on a plane expresses a solid, speech expresses thoughts and truths, characters express 2f>

27

'Monadology', Gerhardt vi. 616, Loemker 648.

Ibid,, Gerhardt vi. 61.7, Loemker 649; 'Consideration on Vital Principles' (1705), Gerhardt vi. 540-1, Loemker 587, 28 Letter to Arnauld, 9 Oct. 1:687, Gerhardt. ii. r 12, Loemker 339,

8o Leibniz and Kant numbers, and an algebraic equation expresses a circle or some other figure. What is common to all these expressions is that we can. pass from a consideration of the properties of the expression to a consideration of the properties expressed. Hence it is clearly not necessary for that which expresses to be similar to the thing expressed, if only a certain analogy is maintained between the properties.29 If the laws of bodies and the laws of souls correspond to each other like two perfectly regulated clocks, then the laws of bodies express the laws of souls—and the laws of souls express the laws of bodies. However, there is an asymmetry here which the mere notion of expression or correspondence fails to capture. The bodies depend on the monads, but the monads do not depend on the bodies. The bodies are nothing over and above the monads, De Voider complained that this means that Leibniz seems to 'eliminate bodies completely'. Leibniz seems to 'place them in appearances, and to substitute for things only forces, not even corporeal forces, but perception and appetite'.30 Leibniz denied it. He replied, 'I don't really eliminate body, but reduce tl to what it w.'31 He says, in a marginal comment on Berkeley's Principles, we don't have to say that matter is Nothing; it suffices to say that it is a phenomenon, like a rainbow; and that it is not a substance but a result of substances,32 Bodies are results of monads, but monads are not results of bodies. Bodies are phenomena: they are not substances, but a result of substances. The reduction of bodies to monads, as Leibniz describes it, seems then to have the following features. There is a correspondence between the properties of bodies and the properties of monads; the realm of monads expresses the realm of bodies, and the realm of bodies expresses the realm of monads; and there is an asymmetry about the correspondence, so that the properties of bodies are the 'results' of the properties of monads, in such a way that bodies are said to be 'reduced' to monads. This is indeed well captured in the idea that bodies supervene on monads, which suggests that Kant's account of Leibniz—on the above interpretation—is accurate enough. One can, as Leibniz puts it, 4pass from a consideration of the properties of the expression to a consideration of the properties of the properties 2

-* * What is an Idea?* (1678), Gerhard t vii. 263—4, translation by Mates, Philosophy of Leibniz, 38. 30 de Volder's words, as quoted by Leibniz in his letter to de Voider, 1704 or 1705, Gerhardt ii. 275, Anew and Garber 181. S1 Letter to de Voider, 1704 or 1705 (italics added), Gerhardt ii. 275, Anew and Garber 181. 32 Mates, Philosophy of faiktiiz, iyy, citing W. Kabitz, Sitzungsberichie der Preusshchen Akademie 4er Wmenschaften, Phil Hist. Klasse 34 (1932), 636; and A, Robinet, 'Leibniz; Lecture du treatise d.e Berkeley,1 Eintkx dephilosophic (11)83), 21.7—23,

Leilmiz and Kimt 81 expressed'. Since the properties of monads entail the properties of bodies, one can pass from a consideration of the properties of monads to the properties of bodies. And since supervenience implies that there is no difference in the properties of bodies without a difference in the properties of monads, one can pass from a consideration, of the properties of bodies to a consideration of the properties of monads, to the following limited extent: where there is a difference in the bodies, one may infer that there is a difference in the properties of things in themselves. Supervenience can be viewed, then, as one particular kind of expression., or mirroring, Kant is right to think that the doctrine of the reduction of bodies to monads is linked to that of the reduction of relations to intrinsic properties: Leibniz says that in the universe, place and position . , . are only relations, resulting from other things . . . Considering the matter more accurately, I saw that they are only mere results, which themselves do not constitute any intrinsic denomination, and thus are only relations which need a fundament from the predicament of quality or an intrinsic accidental denomination.35

Relations, like bodies, are described as 'results', which 'need a fundament' in intrinsic properties. Spatial properties are here described as 'only relations', and dynamical properties, physical forces, are described in the same way. If physical properties are relational, then the reduction of bodies and the reduction of relations go together: the supervenience of bodies on monads, and the supervenience of relations on intrinsic properties, are aspects of the same reduction. As we saw earlier, Kant ascribes to Leibniz a two-tier reduction: of spatial relations to dynamical relations, and then of dynamical, relations to intrinsic properties 0.267/1)323). And Leibniz does indeed take spatial and temporal properties to be less real than dynamical properties; space and time are real 'notperse, but only to the extent that they involve ... the force in created substances'.34 Dynamical properties in turn supervene on the intrinsic properties of monads, physical force being merely 'derivative' force, force that derives from the non-physical 'primitive force' which is an intrinsic property of the monad. Leibniz's reduction of phenomena to things in themselves has thus led ys to his reduction of relations to intrinsic properties. This means it has led us to a topic of considerable dispute amongst Leibniz scholars, and hence considerable well-founded hesitation on the part of any rational newcomer to the field. I attribute to Leibniz a thesis about reducibility which I take to be broadly consistent with Kant's own opinion about Leibniz, and with the 33 34

Couturat , translation from Mates, Philosophy offaibniz, 223.

'Specimen of Dynamics', Gerbartlf M vii, 247, Anew and Garber 130,

82 Leibniz and Kant opinions of a number of recent commentators, though perhaps it is not quite to be found in its entirety in any single one of them.35 On this interpretation, the reducibility of the physical world—the reducibility of the spatial and dynamical relations to which Kant draws attention-—is a consequence of the reducibility of relations in general. So let us turn, then, to the motivation for Leibniz's more general thesis about the reducibility of relations, Leibniz thinks that to exist m to be either a substance or a property of a substance, and since relations are not substances, the only candidate solution is that they are properties. But this creates difficulties. Suppose that David is the father of Solomon. To whom should we ascribe the property in question? David bears the relation 4s father of* to Solomon. David has the relational property of being father-of-Solomon. Solomon bears the relation *is son of to David, Solomon has the relational property of being son-of-David. There are two subjects or substances involved here. So there is a temptation to treat the relation as a property attributable to both, 'Fatherhood'' has, so to speak, one leg in David, and another in Solomon. That is how Leibniz describes the matter in his fifth letter to Clarke, if we were to think of relations as properties, or accidents, we should have an accident in two subjects, with one leg in one and the other in the other; which is contrary to the notion of accidents, Therefore, we must sav that this 35 In brief, I am in agreement with van Cieve, who attributes a supervenience thesis to Leibniz in 'Inner States and Outer Relations'; and I have learnt much from Mates, Reseller, and Parkinson, I follow the majority of commentators in attributing to Leibniz some sort of reducibility thesis, but Ishiguro and Hintikka are notable exceptions to this consensus. In taking reducibility not to be a thesis about the logical equivalence of relational and nonrelational propositions, I disagree with Russell, and agree with van Qeve, Mates, and Rescher. Mates and Rescher see reducibility as requiring merely a one-way entailrnent from intrinsic properties to relational, rather than the logical equivalence required by Russell. A one-way entailment interpretation is weaker than the supervenience interpretation favoured by van Cleve and myself, since it amounts to just the second conjunct of a supervenience thesis about relations. Buroker sees Leibnman reducibility as involving the 'logical priority' of nonrelational properties over relational. This is surely too weak for Leibniz, even for Buroker's own interpretative purposes: as Kant says, a substance must have intrinsic properties., but, need not have relational- a fact which is sufficient for the 'logical priority' of the intrinsic but implies nothing about the reducibility of relations. I see the well-foundedness of phenomena and the founding of the relational in the intrinsic as closely connected themes. In this I agree with Kant, Rescher, and. Mondadori, but not with Adams. See Adams, 'Phenomenalism'; Buroker, Space and Incangruence, 34; van Cleve, 'Inner States and Outer Relations'; Mates, Philosophy of Leibniz, Parkinson, Logic and Reality in Leibniz's Mcta-physics', Mondadori, 'Solipsistic Perception'; Rescher, Phikmophy ofLcibniz\ Jaakko Hintikka, 'Leibniz on Plenitude,, Relations, and the "Reign of Law'"; and. Hide Ishiguro, 'Leibniz's Theory of the Ideality of Relations5, both in Harry Frankfurt, ed., Leibniz (New York; Doubled ay, 1972); Jshiguro, i^ilmiz's Philosophy of Logic and Lemgmge (London; Duckworth, 11)72}; and Bertram! Russell, A Critical Exposition of the Philosophy of Leibniz^ 2nd edn. (London; George Allen & Urswin, 1937).

Leibniz and Kant

83

relation . . . is indeed out of the subjects; but being neither a substance nor an accident, it must be a mere ideal thing, the consideration of which is nevertheless useful,36

It is impossible, according to Leibniz, to have *an accident in two subjects'. In so far as a relation h conceived of as if it were really in. two subjects, it is a 'mere ideal thing*. But the relation none the less has a firm basis in reality,, for it must have 'foundations' in the things to which it is attributed. [TJhere are no purely extrinsic denominations, denominations which have absolutely no foundation in the very thing denominated. For it is necessary that the notion of the subject denominated contain the notion of the predicate. And consequently, whenever the denomination of a thing is changed, there must be a variation in the thing in itself,37 My judgment about relations is that paternity in David is one thing, sonship in Solomon another, but that the relation common to both is a merely mental thing whose basis is the modifications of the individuals.38

The relations must have a basis, a foundation, in the intrinsic modifications of the related individuals.39 How such a notion of reducibility might apply to David and Solomon is not easy to see, but there are simpler cases to hand. Assume for the sake of argument (and contrary to Kant and Leibniz) that height is an intrinsic property. The relation expressed in the proposition * Simmias is taller than Socrates* (to borrow Plato's example) is then reducible not because the proposition is logically equivalent to the conjunction of, say, the propositions 'Sirnniias is six feet tall' and 'Socrates is five feet talP (which it is not),40 It is reducible because it supervenes on the intrinsic properties of Simmias and Socrates: for the relation to hold, Simmias and Socrates must have certain intrinsic properties; and those intrinsic properties must be such that they entail the relation. And that is an accurate description of the case. Both requirements are met. For the relation to hold, Simmias and Socrates must 36

From the letters to Clarke, 1715—16, Gerhardt vii. 401, Anew and Garber 339, 'Primary Truths' (1686), Couturat 520, Anew and Garber 32. 58 Letter to des Bosses, 21 Apr. 1714, Gerhardt ii. 486, Loemker 609, emphasis added. 39 It is not sufficient for reducibility that a relational proposition be implied by a proposition that has only one-place predicates. Relational properties (e.g. is-a-father, and even isa-father-of-Solomon) can be represented as one-place predicates, but they are not for that reason reducible. If they were, fatherhood would be reducible because David is father-ofSolomon, and that, implies that he is father of Solomon. (The proposition having the Fa form implies the proposition having the aRh form.) But this would be a cheat, and in any case is not what Leibniz means, as Mates argues convincingly, Philosophy of Leibniz, 214-15. The reduction must be, not to one-place predicates, but to intrinsic properties. 40 See Plato's Phaedo, trans. Hugh Tredennick, The Colfeet&l Dialogues of Plate, ed. Edith Hamilton and. Huntington Cairns (Princeton, N.J.; Princeton University Press, 10.61), 102. 37

84 Leibniz and Kant be such that they have certain heights; and those heights must entail the relation. Simmias' being six feet tall and Socrates1 being five feet tall is enough for the 'taller than' relation to hold, and therefore enough for the corresponding relational property 'taller than Socrates' to be attributable to Simmi as. It is impossible for Simmias to be six feet tall, Socrates five, and for Simmias to fail to be taller than Socrates. The fact that Simmias is taller than Socrates is nothing over and above the fact that Simmias is six feet tall, and Socrates is five feet tall. Just as the notion of supervenience provides a good way to understand Leibniz's reduction of bodies to monads, so it provides a good way to understand his doctrine of the reducibllity of relations, along the lines just suggested. Many of Leibniz's remarks about relations make it seem clear that the reductibility he has in mind can. be captured in the idea that relations supervene on intrinsic properties. Relations are described as mere 'results', where the results are a kind of non-causal dependence, It is perhaps worth noting—though we cannot put too much weight on this—that Leibniz even seems to use the label of 'supervenienee* to describe the kind of results they are.41 He also says, 'A relation, since it results from a state of things, never comes into being or disappears unless some change is made in its fundament,'42 Here we have something close to the slogan for supervenience: no difference in the supervenient things without difference in the foundational things; no change in relation without change in its fundament. The same point is made in the passage from 'Primary Truths' quoted a moment ago: Svhenever the [extrinsic] denomination of a thing is changed, there must be a variation in the thing in itself.' The 'taller than' relation which Simmias bears to Socrates would never come into being or disappear unless some change were made in its fundament: in the intrinsic properties of Simmias and Socrates. The relation comes into being when, for example, Simmias grows; it would disappear if Socrates should have a belated growth spurt; it did disappear when Socrates ceased to exist, after the unfortunate verdict of the Athenian court, In the above example we have a relation that supervenes on the intrinsic 41

'Relation is an accident; which is in multiple subjects; it is what: results without, any change made in the subjects but supervenes from them', emphasis added. Quoted in Ishiguro, Leibniz's Philosophy^ 71 n. 3, citing Bodemann 74. 'Relatio est accidens quod est in pluribus subjettts estque resultans tantum sen nulia mutatione facta ab iis super venit.' I am unsure how to interpret die idea that relation results 'without any change made in the subjects', which seems at first sight to conflict with the passage quoted immediately hereafter. Perhaps the idea is that the relation can. arise even when no change is made in one of the subjects, as when Socrates, without change 'in himself becomes shorter than Simmias, when Simmias grows. 42 Grua 547, translation from Mates, Philosophy of Leibniz, 223

Leibniz and Kant 85 properties of the relata, taken collectively. The relation in question supervenes on the intrinsic properties of Socrates and Simmias, taken together. It may be that Leibniz holds that all relational properties and relations are reducible in this way: that all are reducible to the intrinsic properties of their relata, taken together-—in other words, that they are what I shall call bilaterally reducible. Restricting our attention to two-place relations, for simplicity's sake, let us say that a relation is bilaterally reducible just in case it bilaterally supervenes on its relata, taken collectively. We can apply Kim's definition of super venience to yield an approximation for-— Bilateral superve niencefor relations: A class of relations A bilaterally supervenes on a class of intrinsic properties B if and only if, necessarily, for every pair of objects x andj, and every relation R in Ay if xRy, then there exist intrinsic properties G and H in B such that x has G, and y has H; and necessarily, for every pair of objects w and z, if w has G and z has H, then wRz. In the discussions ahead I will be talking chiefly about reducibility for relational properties rather than relations. But we can say likewise that a relational property is bilaterally reducible just in case it supervenes on the intrinsic properties of its bearer and of some salient other thing.43 Approximately, applying Kim again, we can have-— Bilateral supervenience for relational properties: A class of relational properties A bilaterally supervenes on a class of intrinsic properties B if and only if, necessarily, for every object xy and every relational property F in. A, if x is F, then there exists a y, and there exist intrinsic properties G and H in $, such that x has G, and j has //; and necessarily, for every pair of objects w and .2, if w has G and z has Hj then w has F. Some relations and relational properties clearly are bilaterally reducible. 43

Kim's definition seems to be. restricted to cases of one-place properties supervening on other one-place properties of the same object; my applications of Kim's definition attempt to give an approximation for more than one-place properties supervening on one-place properties of more than one object. My notions of bilateral and unilateral reducibility are adapted from Keith Campbell's discussion of what he calls bilateral and unilateral relations, Abstract Particulars (Oxford: Bkckwell, 1990). The only place I. have found something like this distinction applied to the task of Leibniz exegesis is in Parkinson, Logic and Reality in Leibniz's Metaphysics, 45 and 147, with the labels 'weak' (bilateral.) and 'strong' (unilateral) redneibility, though, it seems to be in the context of a RusselHan logical equivalence interpretation, of reducibility. For simplicity I have a definition of bilateral reducibility, but strictly one needs a more general notion of multilateral redueibility which, unlike my definitions, would deal with »-place relations for « greater than two.

86 Leibniz and Kant The relation of being taller than someone else seems to be bilaterally reducible, and so does the relational property of being taller than Simmias. Many relations and relational properties, on the other hand,, do not seem to be bilaterally reducible: if Simmias is three feet away from Socrates, that relation ('three feet away from1) and its corresponding relational property ('three-feet-away-from-Socrates1) do not seem to supervene on the intrinsic properties of Simmias and Socrates taken together. But it may well be plausible, all the same, to ascribe to Leibniz the prima facie highly improbable view that all relations and relational properties are bilaterally reducible. Sometimes, though, it seems that Leibniz may have a rather different reducibility thesis in mind. Sometimes Leibniz appears to believe that the facts about one thing considered on its own are sufficient to entail all the relational facts about it, The complete or perfect notion of an individual substance contains all of its predicates, past, present, and future.44 [TJhere are no purely extrinsic denominations, denominations which have absolutely no foundation in the very thing denominated. For it is necessary that the notion of the subject denominated contain the notion of the predicate.45

Indeed, this appears to follow from the Leibnizian mirror thesis, according to which facts about one monad Express' all the facts about the other monads, and hence all the relational facts about the first monad, [TJhe concept of an individual substance includes all its events and all. its denominations, even those which are commonly called extrinsic, that is, those which pertain to it only by virtue of the general connection, of things and from the fact that it expresses the whole universe in its own way.46

The concept of an individual thing is said in these passages to entail all the predicates, including the relational predicates, of that very thing. Moving from predicates to properties, the idea suggested in such passages is that the relations and relational properties of a thing supervene on the intrinsic properties of that very thing. On this second interpretation, to put the point more generally, the relations and relational, properties of things supervene on the intrinsic properties of things, considered not collectively but distrihutively. We need to consider, not both terms of the relation together, but either one of them. This would be a notion of what I shall call unilateral reducibility. A relation 44

'Primary Truths', Couturat 520, Ariew and Garber 32. «Ibid, 46 Letter l:o Arnauld, July 1686, Gerhardt ii. 56, Loemker 337. Parkinson argues that. Lei human reducibility is supposed to follow from the thesis that each monad expresses all the others, Logic and-Reality in Leibniz's Metaphyaks, 147.

Leibniz and Kant 87 Is unilaterally reducible just in case it unilaterally supervenes on the intrinsic properties of the relata, taken distributively; and a relational property is unilaterally reducible just in case it unilaterally supervenes on the intrinsic properties of its bearer—roughly as follows; Unilateral supervenient'? for relations',

A class of relations A unilaterally supervenes on a class of intrinsic properties B if and only if, necessarily, for every pair of objects x and j, and every relation R in A, if xRy, then there exist intrinsic properties G and H in B such that x has G and y has H; and necessarily, for every object IP, if w has G, then there exists a z such that wRz, and if w has H then there exists a z such that zRw. Unila leral mpervmience for re la f tonal properties:

A class of relational properties A unilaterally supervenes on a class of intrinsic properties B if and only if, necessarily, for every object A% and every relational property F'mA, if x is /% then there exists an intrinsic property G in B such that x has G; and necessarily, for every object w, if JP has G, then w has F. We saw that if Sirnmms is taller than Socrates, then the relation ('taller than'), and the relational property ('taller-than-Socrates') are bilaterally reducible, since they supervene on the intrinsic properties of Sirnmias and Socrates, If they were to be unilaterally reducible, by contrast, one would need to be able to tell, from looking at the intrinsic properties of Sirnmias alone, that Sirnmias is taller than Socrates. One would need to be able to tell, from looking at the intrinsic properties of Socrates alone, that Sirnmias is taller than Socrates. Or more accurately, since the point is not epistemologieal but metaphysical, the intrinsic properties of Sirnmias alone would be sufficient to determine that Simmias was taller than Socrates; and likewise for Socrates. The suggestion, for this example, is ludicrous. This relation and relational property are not, on the face of it, unilaterally reducible, since they do not supervene on the intrinsic properties of Simmias on his own, or of Socrates on his own. Note that if they were unilaterally reducible they would a fortiori be bilaterally reducible: if they were to supervene on the intrinsic properties of Simmias alone, then a fortiori they would supervene on the intrinsic properties of Simmias and Socrates. Unilateral reducibility is a stronger thesis than bilateral reducibility; unilateral reducibility implies bilateral reducibility, but not vice versa. Failure of unilateral reducibility does not imply failure of bilateral reducibility, as we see from the example: the 'taller than' relation is not unilaterally reducible, but m nevertheless bilaterally reducible, I'm afraid I cannot produce an illustrative example of any relation or relational property which is plausibly unilaterally reducible.

88 Leibniz and Kant Perhaps this is because there are none, but let us postpone consideration of this question until the next chapter. It may well be plausible, all the same, to ascribe to Leibniz the prima facie even more improbable view that all relations and relational properties are unilaterally reducible, It is not entirely clear at first sight which of these two reducibility theses to attribute to Leibniz: it may be that Leibniz holds each, at different times, or it may be that he conflates them, as indeed some of his commentators seem to. Nicholas Reseller attributes bilateral and unilateral reducibility theses to Leibniz without apparently noticing their distinctness; James Royse attributes a unilateral reducibility thesis to Leibniz and then appears to consider various bilateral theses as candidate elucidations of it (which they cannot be). Some commentators, such as Mates, do not seem to note the unilateral thesis at all. It may be, of course, that Leibniz himself holds both theses without conflating them; since the unilateral thesis implies the bilateral (but not vice versa), if he holds the first he is entitled to hold the second, I shall follow Parkinson in supposing that Leibniz consistently holds both reducibility theses.47 This will assume some importance in the next chapter, but what matters most for our immediate task is that Leibniz held a reducibility thesis of this general form: RedncilnU-ty:

All the relations between things, and the relational properties of things, are reducible to—i.e. supervene on—the intrinsic properties of their relata. I have left this description ambiguous between the unilateral and bilateral possibilities, since it will sometimes be convenient to ignore these complications and have a simple, all-purpose thesis in mind. If all relations supervene on the intrinsic properties of substances, then the dynamical and spatial relations that form the physical world are noth47 Nicholas Reseller, in Philosophy of Leibniz^ gives the two interpretations on 71 and 76 respectively; James Royse, in *Leibn.K and the Reducibility of Relations to Properties', Studia- Leibniti&na ia, No. 2 (1980), 179-204, attributes a unilateral reductibility thesis to Leibniz on p. 181, gives what appear to be various bilateral theses as elucidations thereof from p. 184 onwards; see also Mates, Philosophy of Leibniz, 217—19, Parkinson, who unlike these favours a Russellian equivalence interpretation of reducibility, suggests that when Leibniz says there are no purely extrinsic denominations, he means both (a) when one asserts that A bears a relation to B, the asserted proposition is 'reducible to subject-predicate propositions whose subjects are A and. B respectively' and also (b) in order to assert a relational proposition about A and .5, 'it is in principle enough to know the [non-relational] predicates of A alone or the predicates of B alone' (Logic andReality in Leibniz V/W